GREEN CHEMISTRY:
DENSE CARBON DIOXIDE AND WATER AS ENVIRONMENTALLY
BENIGN REACTION MEDIA
by
Andrew J. Allen
B.S., Chemical Engineering
Texas Tech University
Submitted to the Department of Chemical Engineering in
Partial Fulfillment of the Requirements for the Degree of
MASTER OF SCIENCE IN CHEMICAL ENGINEERING
at the
MASSACUSETTS INSTITUTE OF TECHNOLOGY
August 2004
-
( Massachusetts Institute of Technology 2004,
/
All Rights Reserved
_
Signature of Author:
Department of Chemical Engineering
August 15, 2004
(
Certified by:
Professor Jefferson W. Tester
Thesis Supervisor
Accepted by:
Daniel Blankschtein
Professor of Chemical Engineering
(hairmnn
Cnmmittpt fr
MASSACHUSETTS INSTrlNE
OF TECHNOLOGY
ARCHIVES
SEP 0 2 2004
I
I
LIBRARIES
rnrilntp
tmllieR
2
GREEN CHEMISTRY:
DENSE CARBON DIOXIDE AND WATER AS ENVIRONMENTALLY
BENIGN REACTION MEDIA
by
Andrew J. Allen
To be submitted to the Department of Chemical Engineering August 2004 in partial
fulfillment of the requirements for the Degree of Master of Science in Chemical
Engineering
ABSTRACT
With an ever increasing focus on reducing the environmental impact of solvent
releases on human health and the environment, the replacement of conventional, organic
solvents with alternative compounds that are inherently benign has attracted much
attention in both industry and academia. Supercritical carbon dioxide (scCO2) and water
are two alternative compounds that are of particular interest because they are non-toxic,
non-flammable, readily available, and cheap. Although scCO2 has been successfully used
in industry as a solvent for selective extraction (e.g. extraction of caffeine from coffee
beans), development of scCO 2 as a reaction solvent has been less successful due to its
limited solvation power for many organic reagents of interest. In addition scCO2 has
generally been shown to reduce both the reaction rate and selectivity of many reactions
when compared to conventional solvents. Unlike scCO2, water is known to significantly
accelerate reaction rates and improve selectivities over that obtainable in conventional
solvents. However, most organic compounds are insoluble in water which has limited its
use as a reaction solvent for industrial-scale processes.
In order to replace conventional solvents with scCO2 and/or water, significant
technological advantages resulting from the use of these compounds will have to be
demonstrated. This research attempts to demonstrate some potential advantages of using
scCO2 and scCO2 /water as reaction media for several synthetic transformations of
interest. The Diels-Alder cycloaddition of 9-hydroxymethylanthracene and Nethylmaleimide
was
investigated
in
scCO 2,
and
the
cycloaddition
between
cyclopentadiene and methyl vinyl ketone (MVK) was studied in an scCO2/liquid water
environment. Nitrogen chemistry, specifically the synthesis of nitrogen heterocycles from
amines, was also studied in scCO 2 and scCO 2 /liquid water systems.
The
objective
hydroxymethylanthracene
of
studying
the
with N-ethylmaleimide
Diels-Alder
cycloaddition
of
9-
in scCO 2 was to demonstrate
the
ability of scCO2 to dramatically accelerate the rate of this reaction when compared to
3
conventional solvents. Using spectroscopy to track the disappearance of the 9hydroxymethylanthracene peak, it was found that this reaction proceeds at rates in scCO 2
that are significantly faster than in traditional organic solvents. It was also observed that
the reaction rate constant increased with decreasing density, opposite the trend normally
observed for most reactions conducted in scCO2. On the basis of the low solubility of 9hydroxymethylanthracene in scCO2 and similar results observed in fluorocarbon solvents
(fluorocarbons and scCO 2 are known to behave similarly as solvents), a solvophobic
mechanism was inferred as the cause of the rate acceleration observed for this particular
reaction in scCO 2.
In order to utilize the complementary
solvation powers of scCO 2 and water, a
second Diels-Alder reaction, cyclopentadiene with MVK, was studied in an scCO2/liquid
water mixture. Specifically, the effect of MVK concentration
on the selectivity and
conversion was studied under both silent and sonicated conditions. Power ultrasound was
used as a means to reduce the inherent mass transport limitations that are present in this
biphasic system via the production of emulsions that increase the interfacial contact area
between the two phases. The results of this study demonstrate the ability of power
ultrasound to dramatically increase both the selectivity and conversion over that
obtainable under silent conditions. In agreement with previous studies, it was found that
the ultrasonic effect is most prevalent for systems in which the water-based Hatta number
is greater than one (i.e. an initial MVK concentration of 1.0 mol L- for this system) due
to efficient utilization of the water phase. At lower Hatta numbers (or lower initial MVK
concentrations) the ultrasonic effect is almost negligible as the reaction takes place
mostly in the scCO 2 phase and not in the water phase.
The objective of the final study in this thesis was to develop amine chemistry in
the presence of scCO2 by identifying amines and amine derivatives suitable for
application in a wide range of important synthetic transformations in carbon dioxide
media. The synthesis of nitrogen heterocycles using both hetero Diels-Alder
cycloadditions and Pictet-Spengler cyclizations was investigated. Also studied was
carbamate formation using scCO 2 as both a reagent and solvent. Unfortunately, most of
the results to date are less than encouraging. The hetero Diels-Alder cycloaddition
between benzylamine and 2,3-dimethyl-1,3-butadiene
in scCO 2 failed to produce a
quantitative yield of the desired product; forming instead an unknown oily substance. For
the
Pictet-Spengler
cyclization
in
scCO 2
and
scCO /water
2
systems
of
N-
methoxycarbonyl-3,4-dimethoxyphenethylamine, a carbamate compound, quantitative
yields of the desired product were produced. However, this reaction proceeded much
slower in scCO 2 and scCO/water
2
than in toluene. Due to the slow reaction rate observed
in both alternative media and the inability of scCO 2 or water to dissolve all of the
reactants, the usefulness of either scCO 2 or water as a reaction solvent for this Pictet-
Spengler reaction is questionable. Instead an alternate "one-pot" strategy that uses scCO2
as both a reactant and solvent was pursued. This strategy entails forming the carbamate
from a primary amine and scCO2 prior to injection of additional reagents that would
allow for completion of the Pictet-Spengler reaction in the same reaction vessel.
Unfortunately, attempts to replicate the methods of the Yoshida group for carbamate
generation from scCO2 have failed in our laboratory. We have only been able to obtain
4
about half the yield of the desired product achieved by Yoshida and co-workers with our
remaining product mixture consisting of by-products that the Yoshida group does not find
in their workup.
Thesis Supervisor:
Jefferson W. Tester
H.P. Meissner Professor of Chemical Engineering
5
ACKNOWLEDGEMENTS
There are many people who have made completion of this thesis possible and to
whom I owe a debt of gratitude for the support, encouragement, kindness, and love that
they have given me along the way. Special thanks and acknowledgement go to:
Professor Jeff Tester for welcoming me into his research group with open arms. Your
support and guiding hand for both the completion of this thesis and in helping me decide
on my next step in life are appreciated.
Professor Rick Danheiser for explaining organic chemistry concepts and providing advice
on the experiments conducted in this thesis from the chemist's viewpoint.
Michael Timko and Josh Dunetz both helped introduce me to the CMI project and
answered numerous questions about chemistry, reactors, and anything else that popped
into my head. Mike was especially generous for showing me the ropes as he completed
his thesis and for continuing to answer questions on the phone after he left our lab.
Likewise, Josh was always available for any chemistry questions that I may have had. I
wish him luck as he completes his thesis.
Rocco Ciccolini and Morgan Froling were quite helpful in carrying out investigations as I
focused on writing my thesis. I thank them for all their help and wish them both the best
of luck.
To all the members of the Tester group for lighthearted conversations that helped break
the monotony of research in this dreary basement. Special thanks to Russ Lachance for
answering all of my technical questions about pumps, valves, piping, etc.
I would like to thank my wonderful family, especially my mother and father, for their
love and support that have helped get me to this point in life. Despite the fact that Texas
is a long ways from Cambridge, it feels like you guys are right here with me.
A special thank you goes to Malancha Gupta without whom I would not be finishing this
thesis. You have helped me through some tough times, and I have thoroughly enjoyed
your company during the past year. I promise that I will visit quite often while I am in
law school.
6
-TABLE OF CONTENTSChapter One:
Introduction and Background ............................................................ 10
1.1 Solvents in 'Green Chemistry'..............................................................10
1.2 Water and Carbon Dioxide as Benign Alternative Solvents............................... 12
1.3 References ............................................................
Chapter Two:
Objectives and Approach .............................................................
18
22
2.1 Objectives .............................................................
22
2.2 Method of Approach ............................................................
24
Chapter Three:
Solvophobic Acceleration of a Diels-Alder Reaction in
Supercritical Carbon Dioxide ............................................................
3.1 Background and Objectives .............................................................
26
26
3.2 Experimental Apparatus and Procedures ............................................................30
3.3 Experimental Results .............................................................
38
3.4 Discussion .............................................................
42
3.5 Conclusions and Recommendations .............................................................
53
3.6 References .............................................................
54
Chapter Four:
Water as a Catalyst: Conversion and Selectivity of a Model
Diels-Alder Reaction in Carbon Dioxide/Water Systems ................... 58
4.1 Background and Objectives .............................................................
58
4.2 Experimental Methods ............................................................
61
4.3 Experimental Results .............................................................
65
4.4 Conclusions .............................................................
70
4.5 References .............................................................
70
7
Chapter Five:
Synthesis of Nitrogen-Bearing Compounds in Dense Carbon
Dioxide and/or Water Systems .............................................................
72
5.1 Background and Objectives .............................................................
72
5.2 Experimental Materials and Apparatus ............................................................
77
5.3 Hetero Diels-Alder Cycloaddition .............................................................
79
5.4 Pictet-Spengler Cyclization .
83
............................................................
5.5 Carbamate Generation Using CO2 as a Carbon Source
.............................. 90
5.6 Conclusions and Recommendations .............................................................
95
5.7 References . .......................................................................................................
97
Chapter Six:
Conclusions and Recommendations
..
...................
Chapter Seven: Appendices.............................................................
7.1 Experimental Data: Solvophobic Acceleration
........................................
100
106
106
7.2 Experimental Data: Diels-Alder Conversion and Selectivity .............................. 107
8
-LIST OF FIGURESFigure 3-1:
Figure 3-2:
Figure 3-3:
Figure 3-4:
Simplified schematic of the high-pressure system used to obtain
kinetic data in this study ......................................................
31
Digital picture of the optically-accessible, 3 mL reactor used for
absorption spectroscopy.......................................................
32
Disappearance of the 379 nm peak of 9-hydroxymethylanthracene in
scCO2 during the course of a reaction.......................................................
35
Representative first-order plots for reaction I obtained at 75 °C in
scCO2 at various pressures.......................................................
35
Figure 3-5:
Values of kc measured for reaction I in scCO2 plotted as a function of
pressure and density at various temperatures.................................................40
Figure 3-6:
Solubility of 9-hydroxymethylanthracene
in scCO 2 at 45 °C plotted
as a function of pressure .......................................................
Figure 3-7:
42
(a) Natural logarithms of kxmeasured for reaction I in scCO2 plotted
as a function of pressure. (b) A V+ determined from the slopes of the
curves in panel (a) plotted as a function of reduced temperature .................. 45
Figure 3-8:
Figure 4-1:
Free energy sketches for reaction coordinates representing two
different responses to density .......................................................
47
Schematic of the custom-built, high-pressure reactor used for sonic
experiments (Timko, 2004).......................................................
63
Figure 4-2:
Experimental measurements of (a) the conversion, X, and (b) the
selectivity, S, of the Diels-Alder cycloaddition of methyl vinyl
ketone with cyclopentadiene in carbon dioxide/water systems
plotted versus the initial concentration of dienophile ....................................67
Figure 5-1:
Results of work by the research groups of Leitner (left column) and Albert
(right column) that demonstrate the effect of steric inhibition on
carbamic acid formation from different secondary amines ........................... 76
Figure 5-2:
Figure 5-3:
Digital photograph of the high-pressure view-cell reactor designed
by Weinstein (1998) and used for amine chemistry studies in
this work.......................................................
78
Mechanism for carbamate generation using scCO2 as a carbon source
for the reaction scheme described by the Yoshida group (2000)...................94
9
-LIST OF TABLESTable 1-1:
Estimated annual release and disposal of five common solvents
in the United States in 1997 and 2002 (U.S. EPA) ........................................ 11
Table 3-1:
Rate constants, kc, measured in this study for the reaction of
9-hydroxymethylanthracene
Table 3-2:
and N-ethylmaleimide in scCO 2.................................
39
Rate constants, kc, for the Diels-Alder reaction of
9-hydroxymethylanthracene
and N-ethylmaleimide at
45 °C in hydrocarbon solvents, fluorocarbon solvents, water,
Table 4-1:
and scCO 2 (Qian et al., 2004) ...................................................
52
Characteristic values of rate related processes for dynamics of the
reaction between methyl vinyl ketone and cyclopentadiene
in carbon dioxide/water systems...................................................
68
Table 5-1:
Results of the hetero Diels-Alder cycloaddition in various solvents.............81
Table 5-2:
Results of visual experiments aimed at determining the minimal
ingredients necessary to form the unknown oil-like substance .....................82
Table 5-3:
Yields obtained in various solvents where formaldehyde was used as
the aldehyde source for the Pictet-Spengler cyclization of carbamate ..........87
Table 5-4:
Conversions achieved in toluene and scCO2 where benzaldehyde was
used as the aldehyde source for the Pictet-Spengler cyclization
of carbamate ...................................................
87
Table 5-5:
Yields obtained after complete conversion of the carbamate using
different aldehydes in both toluene and neat (no solvent) systems ...............89
Table 5-6:
Yields obtained from the carbamate generation experiments in
two reactor orientations: vertical and horizontal ...........................................92
10
CHAPTER
1
INTRODUCTION AND BACKGROUND
1.1
Solvents in 'Green Chemistry'
According to the Environmental Protection Agency (EPA), which coined the term
'Green Chemistry', their early definition of the subject is: "'Green Chemistry' is the
utilization of a set of principles that reduces or eliminates the use or generation of
hazardous substances in the design, manufacture, and application of chemical products."
A key aspect of these principles is that they place more emphasis on preventing
environmental problems at the early stages of innovation rather than solving problems
that have already occurred (Anastas and Williamson, 1998). Investigations of chemical
transformations in environmentally-friendly media could potentially lead to greener
chemistry on three levels (Buelow et al., 1998): 1) solvent replacement, 2) better
chemistry (e.g. higher reactivity, selectivity, less energy), and 3) new chemistry (e.g. use
of CO 2 as a carbon source).
The replacement of conventional organic solvents with more environmentallyfriendly media has attracted much attention because solvents are integral to chemical
practices. Solvents are used ubiquitously in a wide range of chemical industries and
applications including roles in reactions, separations, extractions, and cleaning (Nelson,
2000). Worldwide the total solvent market is estimated at over 30 billion pounds per year
(Nelson, 2000), but unfortunately many of these chemicals are still released in significant
11
quantities into the environment despite the proliferation of environmental regulations
during the past three decades (Anastas and Williamson, 1998). As shown in Table 1-1,
the EPA's Toxic Release Inventory (TRI) data indicates that a combined total of over 300
million pounds of methanol, toluene, xylene, methyl vinyl ketone, and acetonitrile were
released and disposed of in 2002. Although the direct impact of solvent releases on
human health and the environment is difficult to assess in monetary terms, the acute and
chronic toxicity and the environmental persistence of many common solvents suggests a
severe impact (Anastas and Williamson, 1998).
The representative reductions in solvent releases shown in Table 1-1 between
1997 and 2002 are largely a result of improvements in solvent recovery and waste stream
treatment. However, the quantities released are still significant and a better long-term
solution would involve a fundamental switch to cleaner solvent technologies. One
solution is to eliminate the solvent, but in cases where this is not possible a second
solution is the replacement of conventional organic solvents with solvents that are
inherently benign.
Table 1-1. Estimated annual release and disposal of five common solvents in the United
States in 1997 and 2002 (U.S. EPA).
Percentage
Solvent
Methanol
1997
(million lbs)
223
2002
(million lbs)
167
Reduction From
1997-2002
25%
Toluene
117
65.7
44%
Xylene
77.9
43.9
44%
Methyl Ethyl Ketone
54.8
28
49%
Acetonitrile
28.4
18.6
35%
12
1.2
Water and Carbon Dioxide as Benign Alternative Solvents
Today there are a wide variety of solvents in industrial use with a considerable
range of physical properties. Besides those listed in Table 1-1 other popular solvents
include carbon disulfide, tetrahydrofuran, acetone, and dichloromethane. Because of their
different properties, all of these solvents are used for many different applications
including recrystallizations,
extractions, cleaning, and coating among others. As a result
of this solvent diversity, it is unreasonable to expect any one 'green' solvent to meet
every industrial need. The motivation of this research is to investigate the use of water
and/or dense carbon dioxide as environmentally benign media for the facilitation of
chemical reactions. Water and carbon dioxide (specifically near- and supercritical,
scCO2) were chosen because they are perhaps two of the most environmentally friendly
solvents as they are non-toxic, non-flammable, readily available, and cheap (Timko,
2004). They were also chosen because they have been the subject of many academic
investigations from which a large knowledge base can be tapped.
Water as a solvent for organic reactions was rediscovered in the eighties after a
long period of time in which it was not considered due to the insolubility of most organic
reactants and the incompatibility of many of the reaction intermediates with water
(Lubineau et al., 1994). Since the early work of both the Breslow group (Breslow et al.,
1983; Rideout and Breslow, 1980) and Grieco et al. (1984; 1983), the effect of water has
been studied in a number of chemical reactions. These investigations have found that
water tends to dramatically increase the rates and/or selectivities of these reactions when
compared to conventional solvents due to a hydrophobic effect that is a consequence of
water's unique hydrogen-bonding network (Lubineau and Auge, 1999).
13
Breslow and Maitra (1984) demonstrated water's ability to dramatically improve
both the reaction rate and selectivity when water was used as a solvent for the DielsAlder cycloaddition of 3-buten-2-one to cyclopentadiene:
0
0
4e
COCH3
Elldo
-
ACOCH3
(1)
Exo
For this reaction, Breslow and Maitra found that the selectivity to the endo product was
improved from 4:1 in octane to 21:1 in water. Water was also found to increase the
second-order rate constant by 700-fold when compared to octane. Meanwhile, Grieco and
Larsen (1986; 1985) also demonstrated water's ability to improve the rate, yield, and
selectivity of a number of organic reactions including hetero Diels-Alder reactions for
cyclocondensations of iminium salts.
Despite these many successes, the aforementioned poor solubility characteristics
of water at moderate temperatures and pressures for most organic reagents is a serious
drawback for its potential use in industry. For example, the solubility of cyclopentadiene
in water at 20
C is only 10 mM. Supercritical water is a good solvent for organic
reagents, but the severe thermal condition and high pressure (Tc > 374 °C, Pc > 240 bar)
is a major deterrent to the use of supercritical water as a solvent in many reactions of
interest. Finally, two other major drawbacks for the use of water in chemical synthesis
(sub- or supercritical) are many chemical processes cannot be performed in protic
solvents like water and many reagents of interest will hydrolyze quickly in the presence
of water (Timko, 2004).
14
As a gas, carbon dioxide is essentially a non-solvent, but above its critical point
(Tc = 31.1 °C, P = 74 bar) the solubility of many compounds in carbon dioxide increases.
Unlike water, carbon dioxide's critical point is relatively mild which makes dense carbon
dioxide a potentially attractive alternative solvent because it is both environmentally
friendly and it has many desirable properties associated with supercritical fluids. A few of
these desirable properties include the large isothermal compressibility characteristic of a
near-critical fluid which allows for large changes in density via small pressure changes.
In theory this ability to change the fluid density allows for manipulation of the reaction
environment (solvation dynamics and equilibrium solubility) through tuning of a number
of physiochemical properties including solvent power, diffusivity, and viscosity. Also,
because the diffusivity and viscosity of supercritical fluids are generally at least an order
of magnitude higher and lower, respectively, than liquid solvents (Lucien and Foster,
1999), interfacial transport limitations can be eliminated in principle through the use of
supercritical fluids. Finally, with supercritical fluids there is the possibility of integrating
both reaction and separation processes which would provide an economic advantage over
conventional synthetic processes that use liquid solvents (Tester et al., 2000).
Industrially, carbon dioxide (liquid, near critical, and supercritical) has been used
successfully in a number of different applications because of its many advantages.
Extraction of caffeine from coffee beans (McHugh and Krukonis, 1994) and the cleaning
of materials with micron-sized features such as those in the microelectronics industry are
a couple of example applications in which carbon dioxide has found widespread use.
Carbon dioxide as a solvent is also attractive to the food and drug industries since
15
residual amounts of CO2 (which should be negligible) do not have any toxic or
undesirable side effects, unlike many conventional solvents.
Despite finding use in a few industrial-scale applications, the high pressures
required for supercritical fluid processing are a major obstacle for more widespread
industrial use of scCO2 . To have an impact on industrial practice, it will be necessary to
demonstrate significant technological and environmental advantages of using scCO2 as a
reaction solvent over conventional compounds. Unfortunately there are two major
drawbacks to scCO 2. Like water, the most important drawback is the low solvation power
of scCO2 for many reagents of interest. In general, low-molecular weight hydrocarbons,
many fluorocarbons (though not highly crystalline fluorocarbons such as conventional
Teflon®-type polymers) (Luna-Barcenas et al., 1998), and many siloxane-based
polymers (Shen et al., 2003) exhibit relatively high (greater than 1 wt%) solubility in
scCO2 at moderate conditions. Most other compounds, including high-molecular weight
hydrocarbons and ionic salts, exhibit extremely low solubilities. The example of benzene,
naphthalene, and anthracene is illustrative of the relationship between molecular weight
and solubility in scCO2. Benzene, the smallest compound of this trio, is almost
completely miscible with scCO2. The solubility of naphthalene is roughly 5 wt% at 35 °C
and 150 bar (Modell et al., 1978), but at the same conditions the solubility of anthracene,
the largest molecule, is less than 5 x 10-3 wt% (Bunker et al., 1997). The second
drawback to using scCO2 as a solvent for reactions is that the rates and selectivities of
most chemical reactions in scCO2 are similar or inferior to those characteristic of
reactions conducted in conventional solvents (Ikushima and Arai, 1999, Renslo et al.,
1997).
16
On an academic scale, efforts to demonstrate the technological advantages of
scCO2 have focused on exploiting its physiochemical properties.
Successful
achievements include using its favorable transport properties to improve reaction rates,
selectivities, and catalyst lifetimes for heterogeneously catalyzed transformations (Baiker,
1999; Subramaniam et al., 2002). The high solubility of gases such as H 2 and CO in
scCO2 has been exploited in a number of hydrogenation (Burk et al., 1995; Goetheer et
al., 2003) and hydroformylation (Davis and Erkey, 2000; Lopez-Castillo et al., 2003)
reactions, sometimes with substantially improved regioselectivities when compared to
conventional solvents (Koch and Leitner, 1998). Near- and supercritical carbon dioxide
have also been used as a reactant (Jessop et al., 1996) and reversible protectant group
(Wittmann et al., 2001).
Without significant technological or economic advantages, environmental benefits
alone are not enough to warrant the replacement of organic solvents with either carbon
dioxide or water. However, one solution to overcome some of the limitations of water
and carbon dioxide is to combine them. Carbon dioxide at conditions near its critical
point is nearly immiscible with water (King, 1992) so that mixtures of the two naturally
form a biphasic system similar to the phase splitting of hydrocarbon/water mixtures.
Because the solvation powers of water and carbon dioxide are complementary, a biphasic
system of the two solvents allows for dissolution of a much wider range of reagents than
is possible in either solvent alone. The accelerative effect of water on chemical kinetics
(the hydrophobic effect) can be accessed more readily by using carbon dioxide to
dissolve the main reagents which tend to be water insoluble. Also in more sophisticated
schemes, such as those involving homogeneous catalysis, segregation of the catalyst in
17
either one of the two phases would allow its reactivity to be accessed and allow for
effective recovery of the catalyst. Despite these possible advantages not all of the
limitations of water and carbon dioxide can be overcome by combining the two
compounds. For instance reactions requiring an aprotic solvent still are not feasible
without protecting agents, and the pH of water in contact with CO2 is effectively buffered
at approximately 3 (Toews et al., 1995) which means that base-catalyzed reactions would
require extremely aggressive buffering.
Another important limitation of a water/carbon dioxide biphasic system is the
imposition of mass transport limitations upon the reaction of interest. In order to remove
mass transport limitations from the water/carbon dioxide system, both emulsions
(Jacobson et al., 1999) and microemulsions (McCarthy et al., 2002) have been
investigated as a means to increase the contact area between the two phases and facilitate
chemical reactions. Overall many of the results are promising as the rates and/or
selectivities of some chemical reactions have been improved over that of the
corresponding conventional solvent system. Most of these studies have relied upon
generating emulsions through the use of surfactants designed for the water/carbon
dioxide interface. However, an alternative approach that has also proven successful is the
use of ultrasound to generate emulsions and thus simplify the system by eliminating the
need for surfactants (Timko, 2004).
In this thesis, both carbon dioxide and carbon dioxide/water
systems were
investigated as reaction solvents. Their effect upon the formation of nitrogen heterocycles
via iminium ions (through hetero Diels-Alder cycloadditions,
Pictet-Spengler
cyclizations, and dipolar cycloadditions) and their role in carbamate synthesis were
18
studied and compared to conventional solvent systems. Although many of the results of
this thesis are encouraging and demonstrate the technological advantages of scCO 2 and
water as reaction solvents, some of the experimental investigations clearly display the
drawbacks of scCO2 as a reaction solvent, particularly when attempting to conduct
nitrogen-bearing chemistry in an scCO 2 environment.
1.3
References
Anastas, P.T. and T.C. Williamson, "Frontiers in Green Chemistry." in Green Chemistry:
Frontiers in Benign Chemical Synthesis and Processes, P.T. Anastas and T.C.
Williamson, Eds., Oxford University Press, Oxford, 1998.
Baiker, A., "Supercritical Fluids in Heterogeneous Catalysis." Chem. Rev. 1999, 99, 453.
Breslow, R. and U. Maitra, "On the Origin of Product Selectivity in Aqueous Diels-Alder
Reactions." Tetrahedron Lett. 1984, 25, 1239.
Breslow, R., U. Maitra and D. Rideout, "Selective Diels-Alder Reactions in Aqueous
Solutions and Suspensions." Tetrahedron Lett. 1983, 24, 1901.
Buelow, S., P. Dell'Orco, D. Morita, D. Pesiri, E. Birnbaum, S. Borkowsky, G. Brown, S.
Feng, L. Luan, D. Morgenstern and W. Tumas, "Recent Advances in Chemistry
and Chemical Processing in Dense Phase Carbon Dioxide at Los Alamos." in
Green Chemistry: Frontiers in Benign Chemical Synthesis and Processes, P.T.
Anastas and T.C. Williamson, Eds., Oxford University Press, Oxford, 1998.
Bunker, C.E., H.W. Rollins, J.R. Gord, and Y. Sun, "Efficient Photodimerization of
Anthracene in Supercritical Carbon Dioxide." J. Org. Chem. 1997, 62, 7324.
Burk, M.J., S. Feng, M.F. Gross and W. Tumas, "Asymmetric Catalytic Hydrogenation
Reactions in Supercritical Carbon Dioxide." J. Am. Chem. Soc. 1995, 117, 8277.
Davis, T. and C. Erkey, "Hydroformylation of Higher Olefins in Supercritical Carbon
Dioxide with HRh(CO)[P(3,5-(CF)3 2-C 6H )3 3] 3." Ind. Eng. Chem. Res. 2000, 39,
3671.
Goetheer, E.L.V., A.W. Verkerk, L.J.P. van den Broeke, E. de Wolf, B.J. Deelman, G.
van Koten and J.T.F. Keurentjes, "Membrane Reactor for Homogeneous Catalysis
in Supercritical Carbon Dioxide." J. Catal. 2003, 219, 126.
19
Grieco, P.A. and S.D. Larsen, "An Intramolecular Immonium Ion Variation of the Diels-
Alder Reaction: Synthesis of ()-dihydrocannivonine." J. Org. Chem. 1986, 51,
3553.
Grieco, P. A., K. Yoshida, and Z. He, "Sodium 6-methoxy-(E,E) -3,5-hexadienoate: A
Useful Diene in the Aqueous Diels-Alder Reaction." Tetrahedron Lett. 1984, 25,
5715.
Grieco, P.A., P. Garner and Z. He, "Micellar Catalysis in the Aqueous Intermolecular
Diels-Alder Reaction: Rate Acceleration and Enhanced Selectivity." Tetrahedron.
Lett. 1983, 24, 1897.
Ikushima, Y. and M. Arai, "Stoichiometric Organic Reactions." in Chemical Synthesis
Using Supercritical Fluids, P.G. Jessop and W. Leitner, Eds., Wiley VCH, New
York, 1999.
Jacobson, G.B., C.T. Lee, R.P. daRocha and K.P. Johnston, "Organic Synthesis in
Water/Carbon Dioxide Emulsions." J. Org. Chem. 1999, 64, 1207.
Jessop, P.G., Y. Hsiao, T. Ikariya and R. Noyori, "Homogeneous Catalysis in
Supercritical Fluids: Hydrogenation of Supercritical Carbon Dioxide to Formic
Acid, Alkyl Formates, and Formamides." J. Am. Chem. Soc. 1996, 118, 344.
King, M.B., A. Mubarak, J.D. Kim and T.R. Bott, "The Mutual Solubilities of Water with
Supercritical Carbon Dioxide." J. Supercrit. Fluids 1992, 5, 297.
Koch, D. and W. Leitner, "Rhodium-Catalyzed Hydroformylation in Supercritical
Carbon Dioxide." J. Am. Chem. Soc. 1998, 120, 13398.
Larsen, S.D. and P.A. Grieco, "Aza Diels-Alder Reactions in Aqueous Solution:
Cyclocondensation of Dienes with Simple Iminium Salts Generated Under
Mannich Conditions." J. Am. Chem. Soc. 1985, 107, 1768.
Lopez-Castillo, Z.K., R. Flores, I. Kani, J.P. Fackler Jr. and A. Akgerman, "Evaluation of
Polymer-Supported Rhodium Catalysts in 1-Octene Hydroformylation in
Supercritical Carbon Dioxide." Ind. Eng. Chem. Res. 2003, 42, 3893.
Lubineau, A. and J. Auge, "Water as a Solvent in Organic Synthesis." Topics in Current
Chemistry, 1999, 206, 1.
Lubineau, A., J. Auge and Y. Queneau, "Water-Promoted Organic Reactions." Synthesis
1994, 8, 741.
Lucien, F.P. and N.R. Foster, "Phase Behavior and Solubility." in Chemical Synthesis
Using Supercritical Fluids, P.G. Jessop and W. Leitner, Eds., Wiley VCH, New
York, 1999.
20
Luna-Barcenas, G., S. Mawson, S. Takishima, J.M. DeSimone, I.C. Sanchez and K.P.
Johnston, "Phase Behavior of Poly(1,1 -dihydroperfluorooctylacrylate) in
Supercritical Carbon Dioxide." Fluid Phase Equilibria 1998, 146, 325.
McCarthy, M., H. Stemmer and W. Leitner, "Catalysis in Inverted Supercritical
CO2/Aqueous Biphasic Media." Green Chem. 2002, 4, 501.
McHugh, M. A. and V.J. Krukonis, Supercritical Fluid Extraction: Principles and
Practice. Butterworth Publishers, Stoneham, MA, 1994.
Modell, M., R.P. de Filippi and V.J. Krukonis, "Regeneration of Activated Carbon with
Supercritical Carbon Dioxide." ACS Annual Meeting, Miami, FL, 1978.
Nelson, W.M, "Choosing Solvents that Promote Green Chemistry." in Green Chemical
Synthesis and Processes, ACS Symposium Series, 767, American Chemical
Society, Washington D.C., 313, 2000.
Renslo, A. R., R.D. Weinstein, J.W. Tester and R.L. Danheiser, "Concerning the
Regiochemical Course of the Diels-Alder Reaction in Supercritical Carbon
Dioxide." J. Org. Chem. 1997, 62, 4530.
Rideout, D.C. and R. Breslow, "Hydrophobic Acceleration of Diels-Alder Reactions." J.
Am. Chem. Soc., 1980, 102, 7816.
Shen, Z., M.A. McHugh, J. Xu, J. Belardi, S. Kilic, A. Mesiano, S. Bane, C. Karnikas, E.
Beckman and R. Enick, "C02-Solubility of Oligomers and Polymers that Contain
the Carbonyl Group." Polymer 2003, 44, 1491.
Subramaniam, B., C.J. Lyon and V. Arunajatesan, "Environmentally Benign Multiphase
Catalysis With Dense Phase Carbon Dioxide." Appl. Catal., B 2002, 37, 279.
Tester, J.W., R.L. Danheiser, R.D. Weinstein, A. Renslo, J.D. Taylor and J.I. Steinfeld,
"Supercritical Fluids as Solvent Replacements in Chemical Synthesis." in Green
Chemical Synthesis and Processes, ACS Symposium Series, 767, American
Chemical Society, Washington D.C., 270, 2000.
Timko, M.T., "Acoustic Emulsions of Liquid, Near-Critical Carbon Dioxide and Water:
Application to Synthetic Chemistry Through Reaction Engineering." Ph.D.
Thesis, Department of Chemical Engineering, Massachusetts Institute of
Technology, Cambridge, MA, 2004.
Toews, K.L., R.M. Shroll and C.M. Wai, "pH-Defining Equilibrium between Water and
Supercritical C02. Influence on SFE of Organics and Metal Chelates." Anal.
Chem. 1995, 67, 4040.
21
U.S. Environmental Protection Agency, Toxic Release Inventory, EPA Offices of
Environmental Information, 2002.
Wittmann, K., W. Wisniewski, R. Mynott, W. Leitner, C.L. Kranemann, T. Rische, P.
Eilbracht, S. Kluwer, J.M. Ernsting and C. Elsevier, "Supercritical Carbon
Dioxide as Solvent and Temporary Protecting Group for Rhodium-Catalyzed
Hydroaminomethylation." Chem.-Eur. J. 2001, 7, 4584.
22
CHAPTER 2
OBJECTIVES AND APPROACH
2.1 Objectives
The overall objective of this study was to investigate the effect of carbon dioxide
and carbon dioxide/water systems on the kinetics of several different types of organic
reactions. Specific objectives in obtaining knowledge about these effects were as follows:
1) Investigate
the solvophobic
acceleration
of a Diels-Alder
reaction in
supercritical carbon dioxide. Most chemical transformations conducted in
supercritical carbon dioxide (scCO2) display reaction rates and selectivities
that are similar or inferior to those characteristic of reactions conducted in
conventional solvents. These studies also indicate that the observed reaction
rate increases with increasing fluid density/pressure. However, there is one
known exception, that of Thompson et al. (1999), in which the observed
reaction rates were faster than those in conventional hydrocarbon solvents and
the reaction rates were also decreasing functions of fluid density/pressure. It
was hypothesized that the observed behavior in the study by Thompson and
co-workers was a result of solvophobic effects.
The approach of this study was to use the knowledge base developed for
Diels-Alder reactions in incompressible solvents in order to gain insight
regarding the molecular-level solvation effects in scCO2. Primarily, our
23
analysis was based on previous research that investigated the use of aqueous
and fluorinated solvents as media for Diels-Alder reactions. In this study we
demonstrate that the Diels-Alder cycloaddition of 9-hydroxymethylanthracene
with N-ethylmaleimide
in scCO 2 shows dramatic rate accelerations
over
conventional solvents and we propose that it is a solvophobic mechanism, a
result of the low solubility of 9-hydroxymethylanthracene,
that leads to the
observed results.
2) Investigate the effect of a biphasic carbon dioxide/water system on the
rate and selectivity of a model Diels-Alder reaction. Combined water and
carbon dioxide systems have attracted attention in green chemistry because
this combination allows one to maintain the environmental benefits of
aqueous reaction media while addressing the solubility limitations of organic
reagents in water. Unfortunately, biphasic systems have large mass transport
limitations. The generation of emulsions via the use of surfactants is one
possible way to overcome these transport limitations. Another strategy is to
generate emulsions through power ultrasound as demonstrated by Timko
(2004). Specifically, this study continued the work of Timko by investigating
the effect of initial methyl vinyl ketone concentration on the selectivity and
conversion of the model Diels-Alder cycloaddition between cyclopentadiene
and methyl vinyl ketone under both silent and sonicated conditions in a dense
carbon dioxide/water system.
3) Study the chemistry of nitrogen-bearing compounds in dense carbon
dioxide and/or water systems. Chemistry of nitrogen-bearing compounds in
24
the presence of carbon dioxide is often difficult due to the tendency of amines
to react with carbon dioxide and form carbamic acid, an undesirable, unstable
side product that inhibits formation of the desired adduct. The intent of this
study was to develop and investigate different reaction classes in the presence
of CO2 and/or water that form nitrogen heterocycles via iminium ions. Hetero
Diels-Alder cycloadditions and Pictet-Spengler cyclizations were the primary
reaction classes studied. Also studied was the generation of carbamate
compounds using scCO 2 as both a reactant and solvent.
2.2 Method of Approach
To accomplish the above objectives, an experimental plan was developed that
consisted of the following sub-projects:
1) Spectroscopic
measurement
of
the
reaction
rate
of
9-
hydroxymethylanthracene with N-ethylmaleimide following the procedures
previously established by Jin Qian and co-workers (2004). This measurement
involved both the collection and analysis of spectroscopic data.
2) Design of 316-stainless steel inserts to occupy the volume of the acoustic
reactor so that a water fraction of 0.85 by volume could be reached while
maintaining the optimal horn-to-interface distance of 2 cm.
3) Measurement of the conversion and selectivity of the well-known Diels-Alder
reaction between methyl vinyl ketone (MVK) and cyclopentadiene at various
initial MVK concentrations in order to demonstrate the benefit of power
ultrasound for a commercially important reaction.
25
4) After completion of the designated reaction time for the hetero Diels-Alder
cycloaddition,
Pictet-Spengler
cyclization,
and
carbamate generation
experiments, the reactor contents were collected and given to Josh Dunetz for
analysis of both yields and conversions of all products and the starting
material, respectively.
5) For the nitrogen chemistry experiments visual observations were conducted in
this lab and all unexpected observations were noted and reported. For runs
demonstrating unexpected behavior, control experiments were conducted as
needed in order to determine the source of this behavior.
26
CHAPTER 3
SOLVOPHOBIC ACCELERTION OF A DIELS-ALDER
REACTION IN SUPERCRITICAL CARBON DIOXIDE
The experiments in this study were conducted in collaboration with Jin Qian,
Michael T. Timko, and Christopher J. Russell at MIT. My involvement included
spectroscopically measuring the rate of the reaction of interest over a span of different
pressures at 75 C. The kinetic data at 45
C and 60 C and the solubility data were
obtained prior to my participation in this analysis. This chapter is a summary of both my
work and the work of the other collaborators involved in this study. More information
can be found in publications by Qian and coauthors (2004; 2003).
3.1 Background and Objectives
Background. Many chemical transformation reactions in supercritical fluids are under
investigation by research groups in the U.S., Europe, and Japan (Tester et al., 2000). The
Diels-Alder cycloaddition, in particular, is a good model reaction for use in supercritical
fluids because it (Timko, 2004):
1) is of commercial importance because it is one of the most important
synthetic pathways for the construction of six-membered rings which are
contained in many important pharmaceuticals and natural products.
2) proceeds
in a controlled
fashion to the desired
temperatures (less than 100 °C).
products
at low
27
3 ) follows bimolecular second-order kinetics; first order with respect to each
of the two reagents.
4') yields a mixture of isomers (either regio- or stereo-) depending on the
choice of reactants, thereby affording an opportunity to study selectivity.
5) allows for the selection of reagents that will ensure that the reaction rates
are fast enough for meaningful kinetic data to be collected in a reasonable
time but slow enough to study in standard batch reactors that require
reactions with half lives greater than several minutes.
Given these advantages it is not surprising that the Diels-Alder reaction has been
the subject of many studies in near-critical and supercritical carbon dioxide (scCO2)
(Glebov et al., 2001; Thompson et al., 1999; Lin and Akgerman, 1999; Renslo et al.,
1997; Weinstein
et al., 1996) and subcritical
(Reaves
and Roberts,
1999) and
supercritical propane (Knutson et al., 1995). All of these studies, with the exception of
one publication (Thompson et al., 1999), have found that the rates of Diels-Alder
reactions are generally slower in scCO2 than the rates observed in conventional
hydrocarbon solvents. These studies also indicate that the observed reaction rate increases
with increasing fluid density/pressure with exceptions having been reported near the
critical point (Thompson et al., 1999; Lin and Akgerman, 1999). Selectivities of the
Diels-Alder reactions in scCO2 are comparable to those observed in hydrocarbon solvents
(Renslo et al., 1997).
In order for scCO 2 to replace conventional compounds, researchers will have to
demonstrate the technological advantages of using scCO2 as a reaction solvent. To date
most research efforts on using scCO2 have focused on exploiting the physiochemical
28
properties of scCO2 . One example includes taking advantage of its favorable transport
properties to improve the rates, selectivities, and catalyst lifetimes of heterogeneously
catalyzed transformations (Subramaniam et al., 2002; Baiker, 1999). Due to the high
solubility
of gases such as H2 and CO in scCO 2, scCO 2 has also been used in
hydrogenation (Goetheer et al., 2003; Burk et al., 1995) and hydroformylation (LopezCastillo et al., 2003; Davis and Erkey, 2000; Koch and Leitner, 1998) reactions to obtain
substantially improved regioselectivities when compared to conventional solvents (Koch
and Leitner, 1998). Carbon dioxide has also been used as a reactant (Jessop et al., 1996)
and reversible protecting group (Wittmann et al., 2001) at conditions near or above its
critical point. Less emphasis, however, has been placed on molecular-level analysis of
solvation effects of scCO 2 on chemical kinetics. Many of the studies of scCO 2 as a
reaction solvent have involved catalyzed systems (Goetheer et al., 2003; Lopez-Castillo
et al., 2003; Wittmann et al., 2001; Davis and Erkey, 2000; Koch and Leitner, 1998;
Jessop et al., 1996; Burk et al., 1995) which are difficult to interpret at a molecular level
because they proceed via complex mechanisms. Meanwhile, previous attempts to predict
or correlate trends for elementary reactions such as the Diels-Alder cycloaddition have
relied upon kinetic- (Weinstein et al., 1996) or thermodynamic- (Thompson et al., 1999;
Reaves and Roberts, 1999; Knutson et al., 1995) fitted parameters without the benefit of
molecular-level analysis.
The approach of this study was to use the knowledge base developed for DielsAlder reactions in incompressible liquid solvents (e.g., hydrocarbons, fluorocarbons,
water) in order to gain insight regarding the molecular-level solvation effects in scCO 2.
Primarily, our analysis was based on previous research that investigated the use of
29
aqueous and fluorinated solvents because observations of rate and selectivity
improvements in these solvents are directly relevant to chemical reactivity in supercritical
fluids. More specifically, since scCO 2 and fluorinated compounds behave similarly as
solvents,
accelerative
mechanisms
in one may be operative
in the other. If this
assumption holds, then the considerable literature of Diels-Alder reactions in
conventional, liquid solvents can be used to interpret observations made in supercritical
fluids and to form a basis for selecting model reactions.
Since the work of Breslow and co-workers (1984; Rideout and Breslow, 1980)
and Grieco et al. (1983), it has been well established that water dramatically improves the
reaction rates and sometimes the selectivities of certain Diels-Alder reactions when
compared to hydrocarbon solvents. Solvophobic interactions and an enhanced hydrogenbonding effect, which preferentially stabilizes the polarized transition state, are
considered the primary contributors to this rate enhancement (Otto and Engberts, 2000).
Recent research has focused on quantifying the relative energetic contributions of these
two effects for a given set of reactants through the use of computer simulations
(Chandrasekhar et al., 2002).
Fluorinated solvents have also been shown to accelerate certain organic reactions
including Diels-Alder cycloadditions (Jenner and Gacem, 2003; Myers and Kumar,
2000), conjugate addition of amines (Jenner and Gacem, 2003), and esterification of
acids and alcohols (Gacem and Jenner, 2003). One of the largest rate improvements is the
50-fold acceleration of the Diels-Alder reaction of 9-hydroxymethylanthracene (diene)
and N-ethylmaleimide (dienophile) when carried out in perfluorohexane rather than
acetonitrile (Myers and Kumar, 2000). Because of the inability of perfluorohexane and
30
acetonitrile to form hydrogen bonds, it has been hypothesized that the origin of the
acceleration in perfluorohexane is entirely solvophobic.
Objectives. The objectives of this phase of our research project was to (1) demonstrate
that the Diels-Alder cycloaddition of 9-hydroxymethylanthracene and N-ethylmaleimide
(reaction I) in scCO2 shows dramatic rate accelerations over conventional solvents and
(2) to understand the mechanism(s) responsible for this acceleration.
nF
0
(I)
0
3.2 Experimental Apparatus and Procedures
Materials.
9-hydroxymethylanthracene, N-ethylmaleimide, and all solvents (HPLC-
grade) were purchased from Sigma Aldrich and used without further purification. Grade
5.5 carbon dioxide (>99.99%) was obtained from BOC and used as received. Water was
deionized to a minimum resistivity of 18.1 MQ cm using a Barnstead Nanopure filtration
system.
Kinetic Measurements.
The simplified experimental setup for the high-pressure
experiments performed in this study is shown in Figure 3-1. Reactions were performed in
an optically accessible reactor, Figure 3-2, designed previously in our research group
(Taylor, 2001). This 316-stainless-steel reactor is cylindrical with a working volume of
approximately 3 cm3 . Sapphire (a-A1 20 3) windows positioned at opposing ends of the
reactor (path length of 2.9 cm) were used for optical access and absorption spectroscopy.
31
Although the reaction mixture could be stirred with a standard Teflon-coated stir bar, a
stir bar was only used in select experiments to confirm that diffusion and free convection
alone were sufficient for good mixing over the time scales of interest because it was
found that the stir bar interfered with spectroscopic measurements. Temperature within
the reactor was monitored to within ±0.5 °C with a T- type thermocouple inserted into the
reaction fluid and maintained using a combination of a PID controller (Omega 9001 CN),
ki'
Figure 3-1. Simplified schematic of the high-pressure system used to obtain kinetic data
in this study. Depicted are (a) cylinder of grade 5.5 C0 2, (b) heat exchanger to liquefy the
gaseous CO2, (c) high-pressure pump (Eldex BBB-4), (d) digital pressure transducer
(Dynisco), (e) HPLC valve (Valco Instruments uw-type), (f) three-way valve that served
as reactor inlet and outlet, (g) high-pressure reactor equipped with ao-A120 3 sapphire
windows for either transmission or scattering spectroscopy, (h) T-type thermocouple, (i)
analog pressure gauge (Matheson 63-3133), (j) source of monochromatic UV radiation,
and (k) spectrophotometer detector.
32
thermal tape (Barnstead), and fiberglass insulating tape. Pressure was measured to within
±1 bar using a standard Bourdon-tube pressure gauge (Matheson 63-3133).
Figure 3-2. Digital picture of the optically-accessible, 3 mL reactor used for absorption
spectroscopy. This reactor was designed previously in our research group (Taylor, 2001).
In order to initiate each reaction a solution injection method was used because
both N-ethylmaleimide and 9-hydroxymethylanthracene are solids at room temperature
and dissolve slowly into scCO2. This method involved dissolving both reactants into
either acetonitrile or acetone and using pressurized CO2 to inject the solution into the
reactor through an HPLC valve (Valco Instruments Co. Inc. uw-type). A second injection
method was used for most of the 75 °C runs in which 10 utLof a solution of both reagents
in either acetone or acetonitrile was directly injected into the reactor prior to sealing the
33
reactor cap at a torque of 100 ft-lbs. This method was adopted because of injection
difficulties, likely a result of a plugged injection loop, that were encountered during the
75 °C experiments.
Visual inspection of the reactor contents indicated one-phase behavior for all
experiments. However, it is sometimes difficult to visually identify the presence of a
small volume of a second phase. Variation of the reactant concentrations served to
confirm that if an undetected second phase were present it did not appreciably affect
kinetic rate measurements. For most of the experimental runs the concentrations of Nethylmaleimide and 9-hydroxymethylanthracene
were 1 x 10-3 mol L-' (1.0 x 10 -4 mole
fraction at a fluid density of 0.4 g cm- 3) and 0.02 mol L-l (2.2 x 10-6 mole fraction at a
fluid density of 0.4 g cm-3), respectively. Although the ratio of dienophile to diene was
varied, it was always kept above 50:1 (in molar units) to ensure pseudo-first-order
kinetics. The mole percentage of co-solvent was approximately 0.2% for most
experiments
and it was always less than 3.5%. Changing
neither
the co-solvent
(acetonitrile or acetone) nor its concentration had a significant effect on the observed
reaction rate within the limits of experimental reproducibility (±5%).
For these studies the reaction temperature was varied from 45 to 75
C and
pressure was varied between 90 and 190 bar. Assuming that the reaction mixture can be
treated as pure carbon dioxide, these conditions correspond to a carbon dioxide density
ranging between 340 and 730 kg m -3 as calculated by an accurate equation of state (Span
and Wagner, 1996). The error introduced by assuming that the reaction mixture is pure
carbon dioxide is minimal given the dilute concentrations of reactants and co-solvent
used in these experiments.
34
The rate of reaction I was measured in situ by monitoring the disappearance of the
9-hydroxymethylanthracene peak at 379 nm using a Varian Cary 50 UV/vis
spectrophotometer.
At this wavelength, interference from the sapphire windows and the
other chemical compounds in the reaction mixture was negligible. Reaction progress was
monitored on the order of 10 h which is approximately
1 to 4 half lives depending on
reaction conditions. Spectra from a representative run are presented in Figure 3-3 while
several representative first-order plots of the logarithm of 9-hydroxymethylanthracene
disappearance over time are shown in Figure 3-4. These logarithmic plots were always
linear with little scatter and the uncertainties in the slopes were less than 4%. The
second-order rate constant was calculated by dividing the slope of the assumed first-order
plot by the known concentration (2% on the basis of its post-reaction recovery) of the
excess dienophile. Control experiments were conducted in which only one of two
reactants was injected into the reactor in order to confirm that the compounds were stable
at the conditions of interest and did not adsorb appreciably to the reactor walls. Drift in
the 379 nm peak of 9-hydroxymethylantrancene was found to be less than ±2%. On the
basis of these three sources of error (scatter in pseudo-first-order plots, reactant
adsorption, and uncertainty in the concentration of excess dienophile), the error in the
measured rate constants is estimated to be less than ±10%.
Analytical Methods. The adduct in the reaction was identified offline by reversed-phase
high-performance liquid chromatography (HPLC; Waters 2690) with tandem mass
spectroscopic (MS/MS; LCQ, Finnigan) detection utilizing an atmospheric pressure
chemical ionization interface. A Zorbax SB-C18 column (4.6 mm x 15 cm) was used
with flash column chromatography (25% ethyl acetate/hexanes). A sample volume of
35
0.30
0.25
UE
Cu
0.20
.I
co
0.15
v
o
.0
.h
oCCL
0
-0
c,
0.10
cu
0.05
1 13
350
360
370
380
390
400
410
420
wavelength (nm)
Figure 3-3. Disappearance of the 379 nm peak of 9-hydroxymethylanthracene in scCO2
during the course of a reaction. Conditions: 75 C, 149 bar (p = 460 kg m 3 ). Spectra
every 4 min are shown for the first hour of a 3 h long run.
0
-0
-0
-0.
,-% -0
-rz
-1
'i
t..
-1
-1
-1
-1
_9
-L..v
0
25
50
75
100
125
150
175
200
time (min)
Figure 3-4. Representative first-order plots for reaction I obtained at 75 °C in scCO 2 at
various pressures: V, 125 bar; , 149 bar; o, 169 bar. [HA] is the instantaneous
concentration of 9-hydroxymethylanthracene and [HA]o is its initial concentration. Solid
lines are the best fits for the data, and the slopes were used to calculate the true second-
order rate constant using the known concentration of excess dienophile.
36
50 L was injected for each analysis. The composition of the binary isocratic mobile
phase was 0.499 acetonitrile/0.499 water/0.002 formic acid. Operating parameters of the
MS/MS system were first auto-optimized in full scan mode and then manually adjusted
by flow injection analysis of each target compound at a concentration of 1 mg L-1. Target
compounds were identified in both full scan mode and selected ion monitoring (SIM)
mode by infusion by matching the retention times and mass spectra with standards. The
adduct was synthesized by refluxing N-ethylmaleimide and 9-hydroxymethylanthracene
in a 9:1 CHC13/CH3 0H solution for 24 h.
For several representative kinetic runs conducted in scCO2, the reactor contents
were collected by depressurizing the reaction mixture through cold acetonitrile. These
post-reaction mixtures were analyzed using the HPLC-MS/MS technique, which
indicated that mass balance closure of the diene was greater than 85% and that the only
product was the Diels-Alder adduct of reaction I.
Solubility Measurements.
Measurements of the solubility of 9-hydroxymethyl-
anthracene in scCO 2 at 45
C and pressures ranging between 70 and 150 bar were
performed by measuring the UV absorbance (at 379 nm using the Varian
spectrophotometer) of samples withdrawn from the reactor. This procedure involved
loading approximately 4 mg of solid 9-hydroxymethylanthracene
into a 25 mL high-
pressure view' cell reactor which is described in detail in Chapter 5. At each pressure, the
supercritical solution was stirred vigorously for 1 h and allowed to settle, unstirred, for
2 h prior to sample collection. A fraction of the mixture (between 10 and 50 pL) was then
withdrawn through an HPLC valve (Valco Instruments uw-type) and slowly discharged
into cold solvent. The precipitated 9-hydroxymethylanthracene
was washed from the loop
37
with additional cold solvent and diluted to a standard volume prior to spectroscopic
analysis.
Kinetic Measurements at Atmospheric Pressure. The rate of the Diels-Alder reaction
of N-ethylmaleimide with 9-hydroxymethylanthracene was measured at atmospheric
pressure in water and acetonitrile over the temperature range 20-50 °C. An Agilent 8532
spectrophotometer equipped with an internal stirring motor and a Peltier heater was used
for the measurements. A 100-fold excess of dienophile was used in all experiments, and
the initial concentration of 9-hydroxymethylanthracene was
x10-3 mol L-1 for
experiments conducted in acetonitrile and 2x10- 5 mol L- 1 for experiments conducted in
water. The disappearance of diene was monitored at 379 nm for at least 2 half lives.
Assumed first-order plots were linear, and the uncertainties in the slopes were less than
1%. The bimolecular rate constant was recovered by dividing the pseudo-first-order
slopes by the known concentration of N-ethylmaleimide. For the water measurements,
concentrated solutions of 9-hydroxymethylanthracene in acetonitrile were used to deliver
the diene. The resulting concentration of acetonitrile was less than 1% on a mole basis.
38
3.3 Experimental Results
Measured Rate Constants.
All 23 measured rate constants (which are referred to as kc
when reported in molar units) for reaction I in scCO2 are listed in Table 3-1 together with
the conditions of the experiment (pressure, temperature, density). As stated previously the
density was determined using an EOS for pure carbon dioxide (Span and Wagner, 1996)
and the error was estimated from the known uncertainties of the pressure and temperature
measurements. At 45 °C and 90 bar, our results indicate that the reaction rate measured in
scCO 2 is nearly 25x faster than that measured in acetonitrile at 45
C and atmospheric
pressure (Myers and Kumar, 2000; Rideout and Breslow, 1980). Considering that other
Diels-Alder
slower
reactions, such as that between cyclopentadiene
in scCO 2 than in many conventional
and ethyl acrylate, are
solvents such as methyl chloride,
tetrahydrofuran, and hexane (Weinstein et al., 1996), this dramatic rate acceleration is
quite remarkable and it demonstrates that scCO 2 as a reaction solvent can provide
significant technological advantages in addition to its many environmentally friendly
characteristics.
39
Table 3-1. Rate constants, k,, measured in this study for the reaction of 9hydroxymethylanthracene and N-ethylmaleimide in scCO2. Reaction conditions are listed
along with the rate data.
Run Number
kc (103
L mol - 1
s-1)a
T (C)
P (bar)
P (kg m 3)
1
25
45
90
340±20
2
23
45
93
3
21
45
97
4
18
45
101
380±20
440± 4
520±30
5
15
45
105
560±20
6
12
45
110
610±20
7
8.0
45
131
698± 8
8
9.6
45
135
710± 8
9
6.8
45
138
717± 7
10
7.4
45
143
729± 7
340±13
11
85
60
85
12
94
60
115
13
64
60
132
14
30
60
152
400±20
520±14
614±10
15
37
60
161
642± 7
16
24
60
166
655± 7
17
24
60
178
684± 6
18
290
75
128
352±
19
230
75
138
406± 9
20
120
75
152
472±30
21
110
75
169
540±
22
76
75
179
572± 7
23
70
75
193
610±22
a Errors
8
8
in the measured rate constants are estimated to be less than 10% based on
the uncertainties in the concentration of N-ethylmaleimide,
errors in the spectroscopic
technique, and adsorption of 9-hydroxymethylanthracene to the reactor walls.
40
Figure 3-5a contains a plot of kc as a function of pressure at a given temperature
while Figure 3-5b is an analogous plot of kc versus the reaction mixture density. These
plots show that at each temperature, the rate constants are roughly linear functions of
pressure and density and the rate constants clearly decrease with both pressure and
density. Also, the magnitudes of the slopes of kc versus either pressure or density increase
smoothly with temperature above the critical point of pure carbon dioxide.
on,\
---
250
250
200
200
L5UU
S.UU
o
E
j
150
c)
--
x
5
150
100
100
50
50
n
80
100
120
140
160
180
200
300
400
500
600
700
800
'3
P (bar)
p (kg m )
Figures 3-5. Values of kc measured for reaction I in scCO2 plotted as a function of
pressure (a) and density (b), respectively, at various temperatures:
, 45 °C; o, 60 °C; Y,
75 C. Solid lines are the best fits for the data but have no physical basis.
To our knowledge the only published report of decreasing rate constants with
increasing pressure/density for a Diels-Alder reaction in scCO2 is that of Thompson et al.
(1999). These authors studied the hetero Diels-Alder reaction of anthracene and 4phenyl-1,2,4-triazoline-3,5-dione
at 40
C (reaction II) which is quite similar to the
model reaction (reaction I) chosen for this study. Therefore, it is not entirely surprising
that the two reactions exhibit similar behavior with pressure and density.
41
Ph
0
Solubilities of 9-Hydroxymethylanthracene
hydroxymethylanthracene
in scCO 2 at 45
0o
in scCO 2 . The measured solubilities of 9C are plotted in Figure 3-6. As shown, the
solubility increases from 1.2x10 - 5 to 1.9x10 -4 mol L- for the pressure range between 70
and 150 bar. Near 75 bar there is a sharp increase in the solubility that is related to the
significant density increase in the vicinity of the critical point of scCO2. The solubility of
9-hydroxymethylanthracene follows a trend similar to that observed for anthracene
(Bunker et al., 1997). Anthracene
hydroxymethylanthracene
is roughly
10
Ox less soluble in scCO
2
than 9-
which can be attributed to two factors: (1) weak specific
interactions between the methanol group of 9-hydroxymethylanthracene and carbon
dioxide and (2) the higher vapor pressure of 9-hydroxymethylanthracene relative to
anthracene due to less efficient packing in the solid state.
42
,,
L~u
I1
I
r
I
I-
I
II
-
F
-·
I
II
I
I
15
-J
E
o1
5
60
80
100
120
140
160
P (bar)
Figure 3-6. Solubility of 9-hydroxymethylanthracene, [HA], in scCO 2 at 45 °C plotted as
a function of pressure. The solid line is intended to guide the eye and no physical basis is
intended (Qian et al., 2004).
3.4 Discussion
The rate of the Diels-Alder reaction of 9-hydroxymethylanthracene and Nethylmaleimide
is accelerated in scCO 2 relative to conventional
solvents, and at all
temperatures for which data are available, the rate constants decrease with increasing
pressure/density.
We propose that the solvophobic effect is the physical cause of these
observations because the solvophobic mechanism is consistent with both the
experimental rate and solubility measurements and it is further supported by activation
volume estimates and comparison to reaction rates in conventional solvents.
Effect of Pressure on Reaction Rates. Transition-state theory is often used by many
researchers to interpret the effect of pressure on kinetic data for reactions conducted in
43
supercritical fluids. Within this context, the pressure dependence of the rate constant is
given by
(r-~r-r------1~~
R(3-1)
;P 'r,
RT
where kx is the bimolecular rate constant expressed in mole fraction units, T and P are the
system absolute temperature and pressure, R is the universal gas constant, x is the
concentration in mole fraction units, and
VI is the apparent activation volume.
Sometimes an overbar is used in conjunction with A VI to specify that it is a partial molar
quantity. The following formula is used to convert between conventional concentration
units, kc, and mole fraction concentration units, kx:
tkc
(p./Mw)
(3-2)
where p is the mixture fluid density (expressed as g L' 1) at the temperature and pressure
of the kinetic measurement and Mw is the average molecular weight of all of the species
in the mixture.
The apparent activation volume is composed of two terms which account for (1)
the formation/breaking of bonds and (2) changes in solvation that accompany the
reaction. For both structured fluids like water and compressible fluids such as those near
their critical point, the apparent activation volume is dominated by changes in the
solvation volumes. Although interpretation of A VI is difficult and rarely provides
mechanistic insight for complicated reactions with multiple steps, molecular-level
information can be obtained in principle from interpretation of A VI for true elementary
reactions such as the Diels-Alder cycloaddition.
44
Using our data and transition-state theory, Figure 3-7a is a plot of the natural
logarithm of rate constants in mole fraction units (i.e. k) as a function of pressure at
different temperatures. The slopes of these curves are essentially linear (R 2 > 0.80), and
the lines are roughly parallel. Equation 1 was used to determine apparent values of AV,
and the results are plotted as a function of reduced temperature (Tr = TIT,) in Figure 3-7b.
The values of a VI for the data set are both large and positive (+350 cm3 mol-'), and the
error bars in Figure 3-7b represent the uncertainties in the slopes of the best-fit lines in
Figure 3-7a.
The studies of Jenner and Gacem (2003) on the effect of pressure on Diels-Alder
reactions, report values of z VIthat range from -27 to -36 cm3 mol'. Using these results,
consideration of only the contributions from bond breaking/formation leads to AzV
estimates of roughly -30 cm3 mo'-1. Since the volume change associated with bond
breaking/formation should not be strongly affected by the solvent, there must be a
significant effect (roughly +380 cm3 mol-') on the apparent values of A VI measured in
this study which arises from solvation.
From Figure 3-7b, proximity to the critical point has only a minimal effect on A VI
since there is only a weak trend in the apparent activation volume in the range of reduced
temperature, 1.06 to 1.23, considered in this study. There may be a slight increasing trend
in A VI with temperature, but uncertainties in the estimated values of A VI are as large as
the total change itself. This observation suggests that critical phenomena arising from
density fluctuations, such as reactant clustering (either solute-solvent or solute-solute), do
not play a role in the observed kinetics (Ellington and Brennecke, 1993).
45
8.0
7.5
7.0
6.5
-C,
6.0
C
5.5
5.0
80
100
120
140
160
180
200
P (bar)
450
400
0
(b)
I
I
350
E
E
300
I~~~~~~~~~~~~
v
++~
-1
250
1
L
200
4
.·
An
, l
_
-------
1.04
1.08
1.12
1.16
1.20
Tr = TITc
Figure 3-7. (a) Natural logarithms of k measured for reaction
function of pressure as suggested by equation 1: , 45 C; o, 60
are the best fits for the data and are related to apparent values of
from the slopes of the curves in panel (a) plotted as a function
I in scCO 2 plotted as a
C; , 75 C. Solid lines
A V. (b) z VI determined
of reduced temperature.
Error bars are based on uncertainties in these slopes (Qian et al., 2004)
46
Several researchers (Thompson et al., 1999; Reaves and Roberts, 1999; Knutson
et al., 1995) have used cubic EOS's to predict the partial molar volume in equation 3-1,
thereby relating measured rate constants (or at least trends in rate constants) to pressure.
This approach usually requires estimates of Tc, P,, the accentricity factor (w), and the
binary interaction parameter (kui) (Tester and Modell, 1997). Since thermodynamic
parameters are well defined only for stable species, it is frequently assumed that the
product and transition state are thermodynamically
equivalent. Also, the interaction
parameters are generally assumed to be zero (Thompson
et al., 1999; Reaves and
Roberts, 1999) or small (Knutson et al., 1995) compared to unity. Naturally these
assumptions introduce some uncertainty and cast suspicion on the use of equation 3-1 as
a predictive tool or its ability to provide physical insight. For example, Thompson et al.
(1999) found that changing the estimated critical temperature of the product (and thus the
transition state) by 20% led to entirely different signs in EOS predictions of A V.
Due to the uncertainty
and suspicions noted above, we propose a different
physical picture to describe the relationship between rate constants and pressure/density
in scCO2 that employs a theory of preferential solvation to explain the observed reaction
rate trends. Figure 3-8 is a schematic of the physical situation in terms of changes in free
energy as a function of reaction coordinate, X. In both panels, there is an increase in free
energy upon combination of the reagents to form the transition state that depicts the
activation free energy for the chemical reaction, zGi. Likewise in both panels, increasing
the fluid density decreases the free energy of the reagents, transition state, and product.
The difference between the panels is the influence of density on the barrier height, A GI.
47
G
G
Reaction Coordinate, X
Reaction Coordinate, X
(a)
(b)
Figure 3-8. Free energy sketches for reaction coordinates representing two different
responses to density. (a) shows preferential solvation of the transition state as density
increases, leading to a net decrease in AG+ with increasing density. In (b) the reactants are
preferentially solvated by increased density, leading to a net increase in AG+ as density
increases (Qian et al., 2004).
In Figure 3-8a, the transition state is preferentially solvated relative to the reactants (Gt
is a decreasing function of density) while the reactants are solvated preferentially relative
to the transition state (G1
is an increasing function of density) in Figure 3-8b. The
reaction rate constant is related to JGI by the Eyring equation:
k,
where kB is Boltzmann's
constant
kBT
exp(-AG*
'RT)
(1.38x10 -2 3 J K - 1) and h is Planck's
(3-3)
constant
(6.6261x10 -3 4 J s). The Eyring equation simply shows that an increase in zJGI leads to an
exponential decrease in k and vice versa. As a result the Eyring equation predicts an
increase in the reaction rate with density for the situation depicted in Figure 3-8a and a
decrease in the reaction rate with density for the opposite situation of Figure 3-8b.
48
Reported kinetic rate measurements of Diels-Alder reactions in scCO2 can be
divided into two groups corresponding to panels (a) and (b), respectively, of Figure 3-8.
The first, more populated group corresponds to Diels-Alder reactions in which the
reactants are rather soluble in scCO2. Many of the most common dienes (isoprene,
cyclopentadiene) and dienophiles (methyl acrylate, ethyl acrylate, maleic anhydride) in
this group are either essentially miscible with scCO2 or at least reasonably soluble. Of the
reagents commonly used in this group, maleic anhydride is the least soluble in scCO 2
with a measured solubility estimated (Glebov et al., 2001) to be 0.2 mol% at 60 °C and
100 bar. This contrasts with the second group, which includes the reagents used in this
study and that of Thompson et al. (1999), in which the solubilities of the dienes are much
lower, on the order of 1x10-3 mol% or less.
Going beyond the physical picture described above, consideration of transitionstate theory provides a more physical interpretation of the observed rate phenomena and
their relationship to reagent solubility. In this context, kc is expressed as
kBT
, = k--7K-C
where K is the transmission coefficient (0
(3-4)
K <1) and KcI is the concentration-based
equilibrium constant for the reaction between the reactants and transition state. For an
elementary reaction such as the Diels-Alder cycloaddition Kc; is defined as
cA-B
(3-5)
CA and CB are the molar concentrations
of the reactants and CAB is the
t
where
"concentration" of the idealized transition state. Following a standard thermodynamic
49
procedure, most of which can be found in the literature (Weinstein, 1998; Weinstein et
al., 1996), equation 3-4 can be written
B exp(
--k
as where JAG
°
lT.
at
(3-6)
is the change in partial molar Gibbs free energy between reactants and
transition state when all species are in their respective standard states, Kxid
is the
continued product of the ideal solution mole fraction solubilities, and Kx,sa; is the
continued product of the actual saturation mole fraction solubilities. By definition KX,idis
not a function of pressure (Prausnitz et al., 1999), and neither is AG, ° provided that the
reference states are defined at a fixed pressure. Therefore, assuming that K is at most a
weak function of pressure, the majority of the pressure dependence in equation 3-6 is due
to Kx,sat.
Using equation 3-6, Figure 3-8a,b can be interpreted in terms of solubility or
solvation. For the Diels-Alder reactions in which the dienes are essentially miscible with
scCO 2 , increases in fluid density will have little effect on the solubility of dienes.
However, the transition state of these reactions is much less solvated in scCO2 than the
reactants due to the larger molecular size of the transition state. It is well-known that as
the molecular size of compounds increase their solubility in scCO 2 decreases. This is
particularly true when molecular size is increased without introduction of new functional
groups. The examples of benzene (which is essentially miscible with scCO2), naphthalene
(which has a characteristic
solubility of approximately
5 wt% in scCO 2 at 35 C and
150 bar (Modell et al., 1978)), and anthracene (the solubility of which is less than
5x10-3 wt% at 35
C and 150 bar (Bunker et al., 1997)) are illustrative. Unlike the
50
reactants used in these types of Diels-Alder reactions, the solvation of the large transition
state, which contains no functional groups not present in the reactants, increases
significantly with fluid density. Therefore, Kx,sat increases with pressure which leads to
an increase in k,.
The
situation
is
reversed
for
dienes
such
as
anthracene
and
9-
hydroxymethylanthracene which are sparingly soluble in scCO2 . The solubilities of these
reactants increase sharply with density as depicted for 9-hydroxymethylanthracene in
Figure 3-6. The solvation of the transition state most likely also increases with fluid
density, but we hypothesize that the effect is less pronounced than for the reactants.
Therefore, for these reactions, Kx,sat decreases with pressure, leading to a decrease in kx.
This hypothesis is supported by the fact that a fraction of the insoluble surface area of the
diene becomes inaccessible to the solvent during the course of the reaction. Due to the
reduced requirement for solvent interactions with the transition state, it is preferentially
solubilized as density is decreased. This is a solvophobic effect analogous to those
reported in water (Chandrasekhar et al., 2002; Otto and Engberts, 2000; Breslow et al.,
1984; Grieco et al., 1983; Rideout and Breslow, 1980) and fluorocarbon solvents (Myers
and Kumar, 2000; Jenner and Gacem, 2003). The large, positive values of A Vf observed
for our rate data are related to changes in the preferential solvation of the transition state
relative to the reactants as pressure decreases rather than true changes in molecular
volumes (of either the reactants/transition state or solvent). Critical phenomena cannot
explain the observed pressure/density behavior since
VI is only a weak function of
reduced temperature. In any event, solute/solute and solute/solvent clustering should be
negligible for Tr > 1.1 (Tucker, 1999).
51
Solvophobic Interactions and Acceleration of Diels-Alder Reactions. The Diels-Alder
reaction between 9-hydroxymethylanthracene and N-ethylmaleimide has been studied
previously in conventional solvents (Myers and Kumar, 2000; Rideout and Breslow,
1980). These studies have clearly implicated the accelerative roles of hydrogen-bonding
and solvophobic interactions. Myers and Kumar (2000) demonstrated that the
bimolecular rate constants for this reaction measured at 45 °C in a range of solvents were
inversely related to the solubilities of the diene in the same solvent, with significant
deviations observed only for reactions conducted in solvents capable of hydrogen bond
donation (e.g., water and trifluoroethanol). Table 3-2 lists available experimental
measurements of the rate constant of reaction I along with the solubility of 9hydroxymethylanthracene
and -z1JGI, which is the change in the activation free energy
relative to that observed in acetonitrile calculated from equation 3-3. Duplicate values of
kc agree to within experimental error with the exception of those obtained in water. In this
study, 1 mol% acetonitrile was used as a delivery solvent for rate studies in water. Most
likely, this small amount of acetonitrile acted as a co-solvent;
thus, reducing the
solvophobic effect and decelerating the reaction rate.
Values of -GI
reported in Table 3-2 range from zero (for acetonitrile, by
definition) to roughly 6 kJ mol-1 (for water). Values of -/zGI
for scCO 2 bridge the gap
between the values reported for fluorocarbon and hydrocarbon solvents. Changes in fluid
density in the supercritical solvent are similar to changes in the chemical identity of
incompressible solvents. Increasing the scCO 2 fluid density from 340 to 730 kg m -3 is
roughly equivalent to changing the solvent from perfluoro-n-butyl ether to 1-butanol. Just
as in conventional solvents, the slowest rate constants are observed in scCO2 under
52
conditions at which 9-hydroxymethylanthracene is most soluble. Naturally, the
accelerative effect of scCO2 is less than that of water since carbon dioxide lacks the
capacity to form hydrogen bonds. Nonetheless, solvophobic reduction of AGI in scCO2 is
significant.
Table 3-2. Rate constants, k, for the Diels-Alder reaction of 9-hydroxymethylanthracene
and N-ethylmaleimide at 45 C in hydrocarbon solvents, fluorocarbon solvents, water,
and scCO2 (Qian et al., 2004).
solvent
k,
(10 5 L mol-1 s -')
-AAG a
(kJ mol -1)
solubilityb
(10 3 mol L)
source
Supercritical Fluid Solvents
scCO 2 (p=
340 g L')
scCO 2 (p=
560 g L-1)
scCO 2 (P=
730 g L')
n-hexane
isooctane
di-n-butyl ether
acetonitrile
2480±250
4.0
0.03
this work
1490±150
3.0
0.13
this work
740±100
1.8
0.18
this work
1.24
Myers
na
20.9
Rideout
29.5
Myers
776±
796±
245±
108±
80
71
16
10
Hydroczarbon Solvents
2.3
2.3
0.9
0.0
Rideout
this work
107± 8
100± 6
methanol
337± 60
1-butanol
344± 25
666± 23
perfluorohexane
perfluoro-2-nbutyl ether
perfluorobenzene
trifluoroethanol
water
1.3
29.9
2.1
na
Fluorocarbon Solvents and Water
4.5
>0.005
5345±308
4562±404
4.3
>0.005
152± 26
841±114
22300±720
22600±700
Myers
Rideout
Rideout
Myers
Myers
0.4
2.4
11.2
Myers
18.1
6.1
0.027
Myers
Myers
Rideout
this work
18400±300
a-AAG is the change in activation free energy relative to that in acetonitrile. bSolubilities
of 9-hydroxymethylanthracene at 45 °Cwhere available.
53
3.5 Conclusions and Recommendations
The Diels-Alder reaction of N-ethylmaleimide with 9-hydroxymethylanthracene
proceeds at rates in supercritical carbon dioxide that are much faster than in traditional
organic solvents. On the basis of the low solubility of 9-hydroxymethylanthracene
in
scCO2, we infer a solvophobic mechanism consistent with that proposed for acceleration
of this reaction in water and fluorocarbon solvents. Strong, negative pressure and density
dependencies of the rate constant were observed that is consistent with a solvophobic
mechanism driven by the positive relationship between fluid density and solute solubility.
The apparent activation volumes are both large and positive (+350 cm3 mol') and only a
weak function of reduced temperature which rules out the influence of clustering
phenomena (solute/solvent or solute/solute) as the cause of the observed accelerations.
Instead, the large activation volumes can be attributed to changes in the solubility of the
reagents relative to that of the transition state with increasing density.
Understanding the solvophobic acceleration of Diels-Alder reactions in scCO2
provides a tool for selection of model reactions to conduct in supercritical fluids.
Typically, reagent selection is based on solubility in scCO2, but our results show that
solvophobic acceleration can provide a second criterion for the choice of reagents. The
similarities of scCO2 with fluorinated solvents (which are more accessible experimentally
than scCO 2) might also be exploited in the future. However, like water, scale-up of
reactions involving sparingly soluble species may be prohibitive for utilizing scCO2 as a
solvent for Diels-Alder reactions and other syntheses.
Future work should be conducted to further understand this solvophobic
acceleration so that the results uncovered by this investigation might be applied to a
54
wider range of organic reactions. Possible focal points include: (1) determination of
specific molecular structures which are expected to interact solvophobically in scCO 2, (2)
further quantification of the solvophobic effect, particularly in the presence of hydrogenbond-donating co-solvents, and (3) development of computational techniques integrating
methods from density functional theory and molecular simulation to further molecularlevel understanding of reactivity in scCO2.
3.6 References
Baiker, A., "Supercritical Fluids in Heterogeneous Catalysis." Chem. Rev. 1999, 99, 453.
Breslow, R. and U. Maitra, "On the Origin of Product Selectivity in Aqueous Diels-Alder
Reactions." Tetrahedron Lett. 1984, 25, 1239.
Bunker, C.E., H.W. Rollins, J.R. Gord, and Y. Sun, "Efficient Photodimerization of
Anthracene in Supercritical Carbon Dioxide." J. Org. Chem. 1997, 62, 7324.
Burk, M.J., S. Feng, M.F. Gross and W. Tumas, "Asymmetric Catalytic Hydrogenation
Reactions in Supercritical Carbon Dioxide." J. Am. Chem. Soc. 1995, 117, 8277.
Chandrasekhar, J., S. Shariffskul and W.L. Jorgensen, "QM/MM Simulations for DielsAlder Reactions in Water: Contribution of Enhanced Hydrogen Bonding at the
Transition State to the Solvent Effect." J. Phys. Chem. B 2002, 106, 8078.
Davis, T. and C. Erkey, "Hydroformylation of Higher Olefins in Supercritical Carbon
Dioxide with HRh(CO)[P(3,5-(CF)3 -C
2 6 H )3 3] 3." Ind. Eng. Chem. Res. 2000, 39,
3671.
Ellington, J.B. and J.F. Brennecke, "Pressure Effect on the Esterification of Phthalic
Anhydride in Supercritical Carbon Dioxide." J. Chem. Soc., Chem. Commun.
1993, 1094.
Gacem, B. and G. Jenner, "Esterification of Sterically Hindered Acids and Alcohols in
Fluorous Media." Tetrahedron Lett. 2003, 44, 1391.
Glebov, E.M., L.G. Krishtopa, V. Stepanov and L.N. Krasnoperov, "Kinetics of a Diels-
Alder Reaction of Maleic Anhydride and Isoprene in Supercritical CO2." J. Phys.
Chem. A 2001, 105, 9427.
55
Goetheer, E.L.V., A.W. Verkerk, L.J.P. van den Broeke, E. de Wolf, B.J. Deelman, G.
van Koten and J.T.F. Keurentjes, "Membrane Reactor for Homogeneous Catalysis
in Supercritical Carbon Dioxide." J. Catal. 2003, 219, 126.
Grieco, P.A., P. Garner and Z. He, "Micellar Catalysis in the Aqueous Intermolecular
Diels-Alder Reaction: Rate Acceleration and Enhanced Selectivity." Tetrahedron.
Lett. 1983, 24, 1897.
Jenner, G. and B. Gacem, "High-Pressure Effect on Organic Reactions in Fluorophobic
Media." J. Phys. Org. Chem. 2003, 16, 265.
Jessop, P.G., Y. Hsiao, T. Ikariya and R. Noyori, "Homogeneous Catalysis in
Supercritical Fluids: Hydrogenation of Supercritical Carbon Dioxide to Formic
Acid, Alkyl Formates, and Formamides." J. Am. Chem. Soc. 1996, 118, 344.
Knutson, B.L., A.K. Dillow, C.L. Liotta and C.A. Eckert, "Kinetics of a Diels-Alder
Reaction in Supercritical Propane." in Innovations in Supercritical Fluids, K.W.
Hutchenson and N.R. Foster, Eds., ACS Symposium Series, 608, American
Chemical Society, Washington, D.C., 166, 1995.
Koch, D. and W. Leitner, "Rhodium-Catalyzed Hydroformylation in Supercritical
Carbon Dioxide." J. Am. Chem. Soc. 1998, 120, 13398.
Lin, B. and A. Akgerman, "Isoprene/Methyl Acrylate Diels-Alder Reaction in
Supercritical Carbon Dioxide." Ind. Eng. Chem. Res. 1999, 38, 4525.
Lopez-Castillo, Z.K., R. Flores, I. Kani, J.P. Fackler Jr. and A. Akgerman, "Evaluation of
Polymer-Supported Rhodium Catalysts in 1-Octene Hydroformylation in
Supercritical Carbon Dioxide." Ind. Eng. Chem. Res. 2003, 42, 3893.
Modell, M., R.P. de Filippi and V.J. Krukonis, "Regeneration of Activated Carbon with
Supercritical Carbon Dioxide." ACS Annual Meeting, Miami, FL, 1978.
Myers, K.E. and K. Kumar, "Fluorophobic Acceleration of Diels-Alder Reactions." J.
Am. Chem. Soc. 2000, 122, 12025.
Otto, S. and J.B.F.N. Engberts, "Diels-Alder Reactions in Water." Pure Appl. Chem.
2000, 72, 1365.
Prausnitz, J.M., R.N. Lichtenthaler and E. Gomes de Azevedo, Molecular
Thermodynamics of Fluid-Phase Equilibria, 3 rd ed. Prentice Hall PTR, Upper
Saddle River, NJ, 1999.
Qian, J., M.T. Timko, A.J. Allen, C.J. Russell, B. Winnik, B. Buckley, J.I. Steinfeld and
J.W. Tester, "Solvophobic Acceleration of Diels-Alder Reactions in Supercritical
Carbon Dioxide." J. Am. Chem. Soc. 2004, 126, 5465.
56
Qian, J., M. Timko, J.W. Tester, J.I. Steinfeld, B. Winnik, B. Buckley and B. Green,
"Synthetic Diels-Alder Chemistry in Liquid and Supercritical Carbon Dioxide
Solvent." ACS Annual Meeting, New Orleans, LA, 2003.
Reaves, J.T. and C.B. Roberts, "Subcritical Solvent Effects on a Parallel Diels-Alder
Reaction Network." Ind. Eng. Chem. Res. 1999, 38, 855.
Renslo, A. R., R.D. Weinstein, J.W. Tester and R.L. Danheiser, "Concerning the
Regiochemical Course of the Diels-Alder Reaction in Supercritical Carbon
Dioxide." J. Org. Chem. 1997, 62, 4530.
Rideout, D.C. and R. Breslow, "Hydrophobic Acceleration of Diels-Alder Reactions." J.
Am. Chem. Soc., 1980, 102, 7816.
Span, R. and W. Wagner, "A New Equation of State for Carbon Dioxide Covering the
Fluid Region from the Triple-Point Temperature to 1100 K at Pressures Up to 800
MPa." J. Phys. Chem. Ref Data 1996, 25, 1509.
Subramaniam, B., C.J. Lyon and V. Arunajatesan, "Environmentally Benign Multiphase
Catalysis With Dense Phase Carbon Dioxide." Appl. Catal., B 2002, 37, 279.
Taylor, J.D., "Hydrothermal Chemistry of Methyl Chloride and MTBE: Experimental
Kinetics and Reaction Pathways." Ph.D. Thesis, Department of Chemical
Engineering, Massachusetts Institute of Technology, Cambridge, MA, 2001.
Tester, J.W., R.L. Danheiser, R.D. Weinstein, A. Renslo, J.D. Taylor and J.I. Steinfeld,
"Supercritical Fluids as Solvent Replacements in Chemical Synthesis." in Green
Chemical Synthesis and Processes, ACS Symposium Series, 767, American
Chemical Society, Washington D.C., 270, 2000.
Tester, J.W. and M. Modell, Thermodynamics and its Applications,
3 rd
ed. Prentice Hall
PTR, Upper Saddle River, NJ, 1997.
Thompson, R.L., R. Glaser, D. Bush, C.L. Liotta and C.A. Eckert, "Rate Variations of a
Hetero-Diels-Alder Reaction in Supercritical Fluid C0 2." Ind. Eng. Chem. Res.
1999, 38, 4220.
Timko, M.T., "Acoustic Emulsions of Liquid, Near-Critical Carbon Dioxide and Water:
Application to Synthetic Chemistry Through Reaction Engineering." Ph.D.
Thesis, Department of Chemical Engineering, Massachusetts Institute of
Technology, Cambridge, MA, 2004.
Tucker, S., "Solvent Density Inhomogeneities in Supercritical Fluids." Chem. Rev. 1999,
99, 391.
57
Weinstein, R.W., "Organic Synthesis in Supercritical Carbon Dioxide." Ph.D. Thesis,
Department of Chemical Engineering, Massachusetts Institute of Technology,
Cambridge, MA, 1998.
Weinstein, R.D., A.R. Renslo, R.L. Danheiser, J.G. Harris and J.W. Tester, "Kinetic
Correlation of Diels-Alder Reactions in Supercritical Carbon Dioxide." J. Phys.
Chem. 1996, 100, 12337.
Wittmann, K., W. Wisniewski, R. Mynott, W. Leitner, C.L. Kranemann, T. Rische, P.
Eilbracht, S. Kluwer, J.M. Ernsting and C. Elsevier, "Supercritical Carbon
Dioxide as Solvent and Temporary Protecting Group for Rhodium-Catalyzed
Hydroaminomethylation." Chem.-Eur. J. 2001, 7, 4584.
58
CHAPTER 4
WATER
AS A CATALYST:
CONVERSION
AND
SELECTIVITY OF A MODEL DIELS-ALDER REACTION IN
CARBON DIOXIDE/WATER SYSTEMS
This chapter describes an experimental investigation of the Diels-Alder reaction
between cyclopentadiene and methyl vinyl ketone conducted in carbon dioxide/water
mixtures. My contribution to this project involved studying the effect of methyl vinyl
ketone concentration on the conversion and selectivity of this reaction under sonicated
conditions. This work was a continuation of the experiments performed previously by
Michael T. Timko in his Ph.D. thesis (2004).
4.1 Background and Objectives
Background. As described in Chapter 1, the work of Breslow and co-workers (Breslow
et al., 1983; Rideout and Breslow, 1980) and Grieco et al. (1984; 1983) in the early
1980s sparked two decades of intense investigation into the effect of water upon the
reactivity of organic compounds. Both researchers independently found that the rates of
certain Diels-Alder reactions were accelerated nearly a 1000-fold when conducted in
aqueous media rather than conventional hydrocarbon solvents. These results changed the
common perception of water as a solvent that was not practical for important synthetic
reactions (due to the limited solubility of most organic compounds) into a solvent that
may be industrially useful because of its well-known environmental compatibility and
recently discovered technological advantages.
59
A recent review by Lubineau and Auge (1999) describes many types of chemical
reactions that demonstrate the advantages, both improved reaction rates and selectivities,
of using water rather than conventional hydrocarbon solvents. Although the role of water
in these reactions is complex and not completely understood, it is agreed that the
hydrophobic effect and enhanced hydrogen bonding of water to highly polarized
transition states dominate the observed accelerative effects of water on chemical kinetics.
The relative contribution of these effects is the subject of current inquiry (Chandrasekhar
et al., 2002).
Despite successful academic research that has exhibited clear technological
advantages of using water over conventional solvents, water has not yet been adopted on
a commercial scale because of the limited solubility of most organic reagents in water. In
effect, this limited solubility places a limit on the maximum production rates that can be
achieved. There are potential engineering solutions to increase production rates but they
all suffer from severe disadvantages. For example, reactors with volumes large enough to
maintain required production rates while ensuring one-phase conditions could be used.
However, the recovery of low concentrations of organic compounds from aqueous
streams would be energy intensive and costly and the low solubilities of the organic
compounds would offset any increases in the production rate achieved by using water
over conventional solvents. Another solution is to deploy a second organic phase in
addition to the water phase or to even use one of the reactants as a second phase. The
problems with this solution are that using a second organic phase would compromise the
environmental advantages of using water as a reaction solvent, and it is possible that the
favorable selectivities anticipated with aqueous-based chemistry will be limited by side
60
reactions occurring in the organic phase (Timko, 2004). Finally, an array of chemistrybased solutions has been proposed to increase the solubility of organic compounds in
water while retaining the other beneficial aspects of aqueous reactivity. These techniques
include the addition of carboxylate groups to common dienes (Grieco et al., 1984), the
creation of glucose-grafted dienes (Lubineau et al., 1994), the use of polar co-solvents
(Blokzilj and Engberts, 1994), and the use of surfactants to solubilize organic reagents
(Diego-Castro and Hailes, 1998). Although many of these methods have been successful,
potential drawbacks include the need for additives like surfactants, limitations on the
range of reagents that can be used, or an increase in the number of chemical steps
required to form the desired product.
Timko (2004) pursued an alternative approach to overcome the solubility
limitations of water. His approach employs a biphasic system in which dense carbon
dioxide is used in conjunction with liquid water in order to maintain the environmental
benefits of aqueous reaction media while addressing the solubility limitations of organic
reagents. In this study the Diels-Alder cycloaddition between cyclopentadiene and methyl
vinyl ketone in biphasic carbon dioxide/water systems was studied. To overcome mass
transport limitations that are present in biphasic systems, power ultrasound was used to
generate emulsions in order to increase the interfacial contact area between phases.
Ultrasound was introduced into the reaction mixture via a titanium sonic probe that was
immersed in the reaction fluid. Experiments were conducted at 30 °C and 80 bar over a
range of water volume fractions with a methyl vinyl ketone concentration of 1 mol L-'. In
non-emulsified, silent systems, it was found that both the selectivity and conversion (after
1 hour) increase with increasing water content. With sonication, both the conversion and
61
selectivity increased dramatically over the silent case. For example, sonication at a water
fraction of 0.60, increased conversion from 30% to 60% and selectivity from 6.5:1 to
greater than 10:1. Furthermore, the experimental results suggest that 2 cm is the optimal
distance between the tip of the sonic horn and the carbon dioxide/water interface. Sonic
enhancement was much less dramatic for horn-to-interface distances which differed from
this optimal value, consistent with a number of published reports on the sonication of
fluids (Laborde et al., 2000; Dahnke and Keil, 1999; McMurray and Wilson, 1999).
Objective. The objective of this study was to extend our work in biphasic systems by
investigating the effect of methyl vinyl ketone concentration on the selectivity and
conversion of the model Diels-Alder cycloaddition between cyclopentadiene and methyl
vinyl ketone (reaction 4-1):
0
COCH3
Endo
Exo
Both silent and sonicated conditions in a dense carbon dioxide/liquid water biphasic
system were evaluated.
4.2 Experimental Methods
Materials. Methyl vinyl ketone (purity > 99%) and HPLC-grade acetone (purity >
99.55%) were obtained from Sigma Aldrich and used as received. Dicyclopentadiene was
also purchased from Sigma Aldrich and cracked at 180 °C to obtain cyclopentadiene with
a purity of roughly 98% as determined by gas chromatography. The freshly prepared
cyclopentadiene was used immediately or stored at -10 °C for up to 12 hours before use.
Water was deionized to a minimum resistivity of 18.1 MQ (Barnstead Nanopure) and
62
used immediately. Grade 5.5 carbon dioxide (purity > 99.9995%) was purchased from
BOC Gases and used without further treatment.
Experimental Conditions. All experiments were conducted in an acoustic reactor
designed previously in our group for use in studies involving high-pressure, ultrasonic
emulsification. This reactor is described in detail elsewhere (Timko, 2004), so only the
salient features are presented here. The high-pressure reactor, as shown in Figure 4-1,
consists of two intersecting cylindrical chambers (i.d. = 1.9 cm) bored in a slab (18.8 cm
x 15.3 cm x 5.1 cm) of 316-stainless steel. Three of the four apertures in the reactor
contain a-A120
3
sapphire windows sealed with stainless steel glands. The pressure seal
was made using a Teflon gasket and the sapphire itself was protected from damage
during tightening by a buna-N rubber gasket (90 durometer, Shore A, provided by New
England Die Cutting, Inc.). The fourth aperture contains a titanium sonic probe (o.d. =
0.80 cm, Sonics and Materials, Inc.) sealed with a stainless steel gland. Teflon and bunaA' gasket were again used to make the pressure seal for this aperture. During experimental
runs the glands were tightened to approximately 90 ft-lbs using a torque wrench.
The experimental conditions of this study replicate those used by Timko (2004).
Reactions were run at 30 °C and 80 bar; an environment which was originally selected by
Timko because the effects of ultrasound were well established under these conditions and
the liquid-like density of carbon dioxide (0.70 g cm-3) at this temperature and pressure
was expected to favor high solubility of the reagents and promote two-phase conditions.
In addition to ultrasound, described below, a recirculation loop was also used to agitate
the contents of the reactor. The recirculation loop was configured so that the water phase
was recalculated through the carbon dioxide phase.
63
Figure 4-1. Schematic of the custom-built, high-pressure reactor used for sonic
experiments. A: stainless steel gland; B: buna-N gasket; C: sapphire window; D: Teflon
gasket; E: 0.25" apertures for machined NPT fittings; F: sonic probe. Pictured on the
right is a digital photograph of the reactor (Timko, 2004).
Ultrasound was introduced directly into the reaction medium using the Branson
sonication unit and titanium probe described previously (Timko, 2004). The operating
frequency of the sonication unit is 20 kHz and the ultrasound was pulsed at a duty of 25%
(based on a 1 s cycle) for reactions studied at a power density of 0.55 W cm -3. Sonication
experiments were only conducted with the horn inserted into the water phase and pointed
up toward the carbon dioxide/water interface. For all experimental runs, the volume of
the water phase was monitored so that the horn tip resided approximately 2 cm from the
interface, conrresponding to the optimal distance determined by Timko (2004).
Experimental Protocol. Selectivity and conversion after 1 hour were measured for the
Diels-Alder reaction between cyclopentadiene and methyl vinyl ketone in a dense carbon
64
dioxide/water system (85% water by volume/15% C0 2). Experimental runs were
conducted both in the presence and absence of emulsifying ultrasound. The initial
concentration of methyl vinyl ketone, [MVK]o, was varied from 0.05 to 1.0 mol L- l.
Larger concentrations of [MVK]o were not used because preliminary experiments
indicated three-phase behavior (water-rich liquid, dense C0 2, and an MVK-rich phase)
for concentrations above 1.0 mol L- of the dienophile in water/carbon dioxide systems at
30 °C and 80 bar.
Due to the design of the acoustic reactor, a 316-stainless steel insert was designed
and fabricated to occupy a portion of the reactor volume so that the experiments could be
conducted with a water fraction of 0.85 while maintaining the 2 cm horn-to-interface
distance. With the horn pointed up, the reactor is designed such that 55 cm3 of the water
phase is needed to obtain the 2 cm distance. As a result, the insert, with a volume of
17.3 cm 3 , was used to displace a portion of the excess CO2 volume thereby reducing the
total reactor volume from 82.5 ± 1.6 cm 3 to 65.2 ± 1.6 cm 3 .
In a typical run a solution of MVK and water was prepared and added to the
reactor. The reactor was then sealed and pressurized to 70 bar with carbon dioxide. For
30 minutes the reactor contents were allowed to come to thermal equilibrium during
which time the mixture was agitated using the recirculation pump (LDC Analytical)
operating at a flow rate of 15 cm3 . Sonication was also commenced at this time for
experiments in which it was used. After thermal equilibrium was achieved, the pressure
of the reactor was increased to 80 bar. The reaction was then initiated by injecting 50 AL
of cyclopentadiene into the reactor through a calibrated sample loop via a standard 6-way
valve (Valco Instruments uw-type) using the recirculating water phase.
65
The reaction was concluded after 1 hour by venting the entire reactor contents
OVRMM2812)
into cold acetone. A high-pressure metering valve (Autoclave Engineers 10
heated with thermal tape (Barnstead) operated at 40 V (as set by a standard variable
transformer) was used to maintain an even flow of depressurizing carbon dioxide and
prevent formation of ice in the vent line. A glass vessel equipped with a frit (10 jpm pore
diameter) was used to improve recovery of volatile components. The cold acetone into
which the reactor contents had been vented was diluted to a standard volume by more
acetone and analyzed by GC (Agilent 6890) using the method described for the
measurement of rate data in pure water in Chapter 7 of Timko (2004). The reactor was
rinsed with acetone and this wash was also analyzed by GC. Uncertainties
in the
calculated selectivity and conversion to a 95% confidence level were based on the error
in the GC calibration which was ± 3%.
4.3 Experimental Results
The results from the variable [MVK]o experiments are presented in Figure 4-2.
The silent experiments were completed prior to my analysis of the sonicated condition.
Under silent conditions decreasing the initial concentration of MVK from 1.0 mol L - to
0.05 mol L-1 increases the selectivity from roughly 6:1 (slightly higher than the 4.4:1
observed in pure carbon dioxide) to nearly 16:1 (much closer to the 20:1 observed in pure
water). Meanwhile, the selectivity is much less sensitive to [MVK]o under sonicated
conditions with an increase of only 13:1 to 16:1 between [MVK]o values of 1.0 mol L and 0.05 mol L-1, respectively.
Figure 4-2 also clearly shows that the conversions
achieved after 1 hour are improved under sonicated conditions, especially when
66
[MVK]o = 1.0 mol L -'. In effect, ultrasonic enhancement of mass transport rates allows
water-like selectivities and conversions to be obtained even when [MVK]o = 1.0 mol L-1.
67
4
f%
I
I.U
I
.
I
I
.
I
I
.
I
I
I
I
.
.
.
.
.
.
I,,
.
.
,
.
.
,
.
.
.
.
,
.
.
.
.
0.8
X 0.6
0.4
I
1
Sonicated
-0-. . . . I . . . . . . . . . I . .
0.0
0.2
0.4
0.6
,,
U.L
0.8
1.0
1.2
[MVK]0 (mol L 1 )
18
.
.
. I
. . , .,
. . I ....
I I. . I
I
I
16
14
0x
0 12
0
-
a)
II
10
8
6
4
0.0
d
; onlcated
I
0.2
0.4
0.6
0.8
1.0
1.2
[MVK]0 (mol L 1 )
Figure 4-2. Experimental measurements of (a) the conversion, X, and (b) the selectivity,
S, of the Diels-Alder cycloaddition of methyl vinyl ketone with cyclopentadiene in
carbon dioxide/water systems plotted versus the initial concentration of dienophile. For
the sonicated condition, the horn-to-interface distance is 2 cm. Conditions: 30 °C, 80 bar,
85% water by volume, 1 hr reaction time. Ultrasound: 20 kHz ultrasound, 25% duty (on
a 1 s cycle), 0.55 W cm -3 .
68
Following the analysis of Timko (2004), the observed selectivity behavior can be
interpreted in terms of the relative rates of the processes described in Table 4-1. Table 4-1
lists estimated mass transport coefficients, rate constants, and Hatta numbers, defined as
/ 1 2 / k,
(krxnDA)
in both phases for the extremes in [MVK]o. The mass transport
coefficients listed in Table 4-1 are based on the transport measurements of benzaldehyde
described by Timko and therefore do not take into account transport to water droplets as
they recirculate through the carbon dioxide phase. Thus the mass transport coefficients in
Table 4-1 underestimate the total mass transport rates between the two phases. The Hatta
numbers indicate that: 1) mass transport in the water film is slightly enhanced by reaction
for [MVK]o = .0 mol L-1 as the water-based Hatta number equals 2 in this case; 2) the
carbon dioxide phase is well mixed, as the Hatta number is always much less than unity
in this phase. Water-based Dimkohler numbers are less than unity (< 1 x 10-2).
Table 4-1. Characteristic values of rate related processes for dynamics of the reaction
between methyl vinyl ketone and cyclopentadiene in carbon dioxide/water systems
(Timko, 2004).
Rate process
pseudo-first order reaction ratea (s'- )
mass transport coefficient (cm s -1)
Hatta numberb
reaction enhanced transport coefficient
[MVK]o = 1.0 mol LH 20
C0 2
0.04
2 x 10-5
3x
10
3 x 103-
-4
[MVKjo = 0.05 mol L H 20
CO 2
3 x 10 - 3
1 X 10 -6
3x
10-
4
3 x 10-3
(H2 0 film)
(CO2 film)
(H 2 0 film)
(CO2 film)
2
0.015
0.5
4 x 10- 3
7 x 10- 4
3 x 10- 3
3 x 10-4
3 x 10- 3
(cm s')
overall mass transport coefficient (cm
S-l), kmt kw / Kcw
with Kcw(cyclopentadiene)
6 x 10 -6
3 x 10 -6
1x
6 x 10-7
=100
mass transport coefficient x specific
106
surface areac (s - 1)
overall rate of disappearance of cyclopentadiene i.e., k + k or kc + kc
1/2 / kw or yc
1 /2 / kc
b Hatta number, yw = (krxnDAW)
= krxn DAC)
a for the
c assuming
that the specific interfacial area = 20 cm2 / 100 cm3 = 0.2 cm-1.
69
Using the data in Table 4-1, when [MVK]o = 1.0 mol L-' the pseudo-first-order
reaction rate constant in the carbon dioxide phase is 2 x 10-5 s -1 while the overall rate is
1 x 10-6 s-l . The 20-fold difference between the two rates implies that the reaction occurs
primarily in the carbon dioxide phase before the cyclopentadiene can be transported to
the water phase. This marked difference in rate supports the experimental observation
that the selectivity at [MVK]o = 1.0 mol L-' is much closer to the selectivity observed in
pure carbon dioxide that that observed in pure water. In contrast, when [MVK]o =
0.05 mol L- the reaction rate constant in carbon dioxide is 1 x 10-6 s -1 while the effective
transport-limited
rate constant to the water phase is 6 x 10- 7 s -1, less than a two-fold
difference. This leads to the conclusion that when [MVK]o = 0.05 mol L-1 transport of
cyclopentadiene to the water phase is sufficiently fast compared to reaction in the bulk
carbon dioxide that a substantial fraction of the water phase is utilized, thus supporting
the observation that the selectivity approaches the values reported in pure water at lower
concentrations of methyl vinyl ketone. Meanwhile, the trends observed under sonicated
conditions are consistent with the findings of Timko (2004) for a biphasic system with
small water-based Hatta numbers where a much smaller acoustic effect for systems with
low Hatta numbers (Hatta < 0.5) when compared to systems with Hatta > 1 was reported.
As Table 4-1 shows, the Hatta number in the water phase is approximately 0.5 when
[MVK]o = 0.05 mol L- 1, which supports the observation
that sonication does not
dramatically improve either the selectivity or the conversion at this condition. In contrast,
the significant enhancement in selectivity and conversion that is observed under
sonicated conditions when [MVK]o = 1.0 mol L-' is expected as the Hatta number in the
water phase is equal to 2.
70
4.4 Conclusions
A set of experiments were conducted at a water volume fraction of 0.85 while
maintaining the optimal horn-to-interface distance of 2 cm and varying the initial
concentration of methyl vinyl ketone from 0.05 to 1.0 mol L - . Under sonicated
conditions the most dramatic improvements were seen when the initial methyl vinyl
ketone concentration was 1 mol L- . At this concentration the endo:exo selectivity was
nearly 13:1 and the conversion after one hour was greater than 90%. These results
compare favorably to that obtained silently (6:1 selectivity, 70% conversion) or expected
in pure carbon dioxide (4.5:1 selectivity, 30% conversion).
As the concentration
of
methyl vinyl ketone was decreased from 1.0 to 0.05 mol L- 1, the selectivities and
conversions obtained under sonicated conditions approached those obtained without
ultrasonic emulsification (16:1 selectivity, 50% conversion). In general, for water/CO 2
systems in which the water-based Hatta number is greater than one, sonication increases
mass transport rates in this biphasic system, thus allowing both the selectivity and
conversion to approach that of a water-phase reaction without directly encountering
solubility limitations of the reactants in water.
4.5 References
Blokzilj, W. and J.B.F.N. Engberts, "Enforced Hydrophobic Interactions and HydrogenBonding in the Acceleration of Diels-Alder Reactions in Water." in Structure and
Reactivity in Water, ACS Symposium Series, 568, American Chemical Society,
Washington D.C., 303, 1994.
Breslow, R., U. Maitra and D. Rideout, "Selective Diels-Alder Reactions in Aqueous
Solutions and Suspensions." Tetrahedron Lett. 1983, 24, 1901.
71
Chandrasekhar, J., S. Shariffskul and W.L. Jorgensen, "QM/MM Simulations for DielsAlder Reactions in Water: Contribution of Enhanced Hydrogen Bonding at the
Transition State to the Solvent Effect." J. Phys. Chem. B 2002, 106, 8078.
Dahnke, S. and F.J. Keil, "Modeling of Linear Pressure Fields in Sonochemical
Reactors Considering an Inhomogeneous Density Distribution of Cavitation
Bubbles." Chem. Eng. Sci. 1999, 54, 2865.
Diego-Castro, M.J. and H.C. Hailes, "Studies on the Use of Surfactants in Aqueous
Diels-Alder Reactions." Tetrahedron Lett., 1998, 39, 2211.
Grieco, P. A., K. Yoshida, and Z. He, "Sodium 6-methoxy-(E,E) -3,5-hexadienoate: A
Useful Diene in the Aqueous Diels-Alder Reaction." Tetrahedron Lett. 1984, 25,
5715.
Grieco, P.A., P. Garner and Z. He, "Micellar Catalysis in the Aqueous Intermolecular
Diels-Alder Reaction: Rate Acceleration and Enhanced Selectivity." Tetrahedron.
Lett. 1983, 24, 1897.
Laborde, J.L., A. Hita, J.P. Caltagirone and A. Gerard, "Fluid Dynamics Phenomena
Induced by Power Ultrasounds." Ultrason. 2000, 38, 297.
Lubineau, A. and J. Auge, "Water as a Solvent in Organic Synthesis." Topics in Current
Chemistry, 1999, 206, 1.
Lubineau, A., H. Bienaym6, Y. Queneau and M.C. Scherrmann, "Aqueous
Cycloadditions Using Glyco-Organic Substrates - Thermodynamics of the
Reaction." New J. Chem. 1994, 18, 279.
McMurray, H.N. and B.P. Wilson, "Mechanistic and Spatial Study of Ultrasonically
Induced Luminol Chemiluminescence." J. Phys. Chem. A. 1999, 103, 3955.
Rideout, D.C. and R. Breslow, "Hydrophobic Acceleration of Diels-Alder Reactions." J.
Am. Chem. Soc., 1980, 102, 7816.
Timko, M.T., "Acoustic Emulsions of Liquid, Near-Critical Carbon Dioxide and Water:
Application to Synthetic Chemistry Through Reaction Engineering." Ph.D.
Thesis, Department of Chemical Engineering, Massachusetts Institute of
Technology, Cambridge, MA, 2004.
72
CHAPTER 5
SYNTHESIS OF NITROGEN-BEARING COMPOUNDS IN
DENSE CARBON DIOXIDE AND/OR WATER SYSTEMS
The work in this chapter was conducted as part of the CMI project which is a
collaborative project between the following research groups: Professor Jeff Tester (MIT),
Professor Rick Danheiser (MIT), and Professor Andy Holmes (Cambridge University,
UK). At MIT our lab worked closely with Josh Dunetz of Professor Danheiser's lab. The
overall objective of the CMI project is to investigate the application of water and carbon
dioxide as "environmentally friendly" media for synthetic, organic reactions involving
different types of amines.
5.1 Background and Objectives
Background.
The synthesis of organic compounds in the absence of regulated or
potentially toxic solvents is an important industrial goal, and is the motivation
for
development of alternative reaction media. Supercritical carbon dioxide has recently
attracted considerable attention (Qian et al., 2004; Leitner 2002; Glebov et al., 2001;
Thompson et al., 1999) as it is non-toxic, inexpensive and universally available. Because
supercritical fluids offer the ability to vary density and solvent power through small
changes in pressure and temperature, separations of reactants, products, and catalysts are
facilitated. The many favorable and tuneable properties of near-critical and supercritical
carbon dioxide (scCO2) make it attractive for use on an industrial scale.
73
Besides scCO 2, water has attracted attention as a potential green solvent due to the
work of Breslow and co-workers (1983; Rideout and Breslow, 1980) and Grieco et al.
(1984; 1983) that demonstrated water's ability to accelerate the rate of certain DielsAlder reactions over a 1000-fold when compared to conventional organic compounds. A
recent review by Lubineau and Auge (1999) describes many types of chemical reactions
that demonstrate the advantages, both accelerated reaction rates and improved
selectivities (regio- and stereo-), of using water as a reaction solvent rather than
conventional hydrocarbon solvents. Although the role of water in these reactions is
complex and not completely understood, it is agreed that the hydrophobic effect and
enhanced hydrogen bonding of water to highly polarized transition states (Otto and
Engberts, 2000) dominate the observed accelerative effects of water on chemical kinetics.
The relative contribution of these effects is the subject of current inquiry (Chandrasekhar
et al., 2002).
Unfortunately the practical application of both water and scCO2 as reaction media
has been limited by the poor solvent power they exhibit for many reactants and reagents
of interest. Although scCO2 will dissolve many non-polar compounds of low molecular
mass, many catalysts, substrates, and reagents have limited solubility in this medium.
Likewise, most organic reagents display limited solubility in water. The combination of
water with scCO2 to form a biphasic system is one possible solution to overcome some of
the solubility limitations of both solvents because their solvation powers are
complementary. In addition, this particular biphasic system allows the accelerative effect
of water on chemical kinetics (the hydrophobic effect) to be accessed more readily by
using carbon dioxide to dissolve the main reagents which tend to be water insoluble. A
74
limitation of using a biphasic system is the inherent, interphase mass transport resistance
that may influence the kinetics of the reaction of interest. One strategy to overcome these
potential limitations is the generation of water/CO2 emulsions which reduce mass
transport limitations through the creation of more interfacial surface area. Emulsions can
be created using either surfactants (Jacobson et al., 1999; McCarthy et al., 2002) or power
ultrasound (Timko, 2004).
To this point, the research in this thesis has focused on Diels-Alder reactions that
proceed in a controlled fashion to the desired products and have only involved the
formation and/or breaking of carbon-carbon bonds. Nitrogen-bearing chemistry,
however, is also of interest as nitrogen heterocycles represent a class of compounds that
is extensively used in the pharmaceutical,
agrochemical,
and electronics industries.
Pharmaceutically important compounds include among others: methopholine, yohimbine,
vincamine, salsolinol, and debrisoquine. Some synthetic transformations important for
the formation of these nitrogen heterocycles are:
*
palladium-catalysed amination reactions
·
dipolar cycloadditions including nitrile oxide and azomethine ylide
cycloadditions
*
hetero Diels-Alder reactions of iminium ions
* Mannich and Pictet-Spengler reactions
* enantioselective organocatalytic carbon-carbon bond forming
reactions
* palladium-catalysed coupling reactions on solid phase supports.
Although the chemistry of nitrogen-bearing compounds in conventional solvents
is well documented
(Yokoyama et al., 1999; Gothelf and Jorgensen, 1998; Cox and
75
Cook, 1995; Larson and Grieco, 1985; Kumar et al., 1982), this chemistry is more
difficult in the presence of carbon dioxide as amines can react reversibly with CO2 to
form carbamic acid or alkylammonium
carbamic acid salts (Belli Dell'Amico
et al.,
2003). Both of these compounds are unstable and can lead to the production of undesired
by-products. These reactions are shown below where reaction 5-la is a generic scheme
for carbamic acid formation and reaction 5-lb is the generic scheme for alkylammonium
carbamic acid salt formation.
0
CO2
.
.....
K-NM 2
R
' 'N'
.[I
OH
(5-1a)
H
C0 2
2
R-NH
2
0
R
N
K0 . OH3®N-R
(5-lb)
H
Carbamic acid formation from various primary and secondary amines has been
documented using both visual and NMR techniques (Kainz et al., 1999; Fischer et al.,
2003). Leitner and co-workers conducted visual experiments involving the secondary
amines n-ethyl-n-benzylaniline
and n-methyl-n-benzylaniline
in the presence of CO 2. In
their studies the Leitner group found that n-ethyl-n-benzylaniline formed a white solid
that was attributed to carbamic acid formation. In contrast the Leitner group observed that
n-methyl-n-benzylaniline, which has a bulkier substituent group attached to the nitrogen
atom, did not form carbamic acid as no white solid was observed. Albert and co-workers
conducted experiments that analyzed the amine/carbamic acid equilibrium in scCO 2 via
NMR at 50 "C and 80-200 bar. These experiments focused on various benzylamine
76
derivatives. As with the Leitner group, Allbert and co-workers found that carbamic acid
formation from secondary amines could be prevented by using bulkier substituent groups
to sterically hinder carbon dioxide's ac:cess to the reactive nitrogen atom. More
specifically, attachment of an isopropyl o:r butyl group to the nitrogen atom prevented
carbamic acid formation whereas a methyl group did not. A summary of both the Leitner
and Albert groups' results are shown in the Figure 5-1:
NVR in scCO2
H
(50 °C, 80-200 bar)
carbamic acid
CH3
white solid obseved
¢-N-
(carbanic acid)
no carbamic acid
-'~U
CH3
_
no carbaric acid observed
N
k
no carbaic
no cart
acid
Figure 5-1. Results of work by the research groups of Leitner (left column) and Albert
(right column) that demonstrate the effect of steric inhibition on carbamic acid formation
from different secondary amines.
Supercritical carbon dioxide has also been used in applications where it is both a
reagent and a solvent (Salvatore et al., 2002; Selva et al., 2002; Yoshida et al., 2000;
McGhee et al., 1995). In these studies, scCO 2 was used as a replacement for phosgene to
generate carbamate products from primary and secondary amines at relatively mild
conditions. Some of the general schemes that have been proposed are shown in reactions
5-2 (Yoshida et al., 2000) and 5-3 (Selva et al., 2002):
O
NH
12
R
+
R3 --X
K2 C0 3 , Bu4 NBr
scCO 2
R1
N
12
R
LoR
OR
3
(5-2)
77
0
R1-tNH2
0
R N
scCO2
+
MeO
N
OMe
OMe(5-3)
OMe
H
These carbamate reactions are of interest because they represent a "green
chemistry" methodology to generate a protecting group in situ for the nitrogen atom that
can be used in synthesis reactions of nitrogen heterocycles. The work of Venkov and
Lukanov (1992), shown in reaction 5-4, demonstrates the use of carbamates as a reagent
in Pictet-Spengler reactions involving n-acyliminium ions:
1.5 equiv (HCHO)n
0.02 equiv TsOH
H-H20
MeO
'
x
'
H'
'CO 2Me
toluene (0.5 M)
110°C, 3h
(5-4)
M"
Me
CN2Me
O
75%
Here the -CO 2Me substituent is used as a protecting group to prevent carbamic acid
generation and to produce high yields of the desired product.
Objectives. The objective of this phase of our research was to develop amine chemistry
in the presence of scCO2 by identifying amines and amine derivatives suitable for
application in a wide range of important synthetic transformations in carbon dioxide
media. The synthesis of nitrogen heterocycles using both hetero Diels-Alder
cycloadditions and Pictet-Spengler cyclizations was investigated. Also studied was
carbamate formation using scCO 2 as both a reagent and solvent.
5.2 Experimental Materials and Apparatus
Materials. With the exception of the carbamate compound, all of the reagents, catalysts,
and solvents used in these experiments were purchased and used as received from Sigma
78
Aldrich. The carbamate was synthesized by Josh Dunetz by reacting 3,4dimethoxyphenethylamine
with
methyl
chloroformate
and
triethylamine
in
dichloromethane. Grade 5.5 carbon dioxide (>99.99%) was obtained from BOC and used
as received. DIeionizedwater as received from MIT's water supply was further deionized
to a minimum resistivity of 18.1 MQ cm using a Barnstead Nanopure filtration system.
Apparatus.
All experiments were performed in the view-cell high-pressure reactor
designed and described in detail by Weinstein (1998; Renslo et al., 1997); thus only the
basic features of the system are presented here. The reactor, Figure 5-2, was fabricated
from a 6.4 cm x 6.4 cm x 12.7 cm block of 316-stainless steel. A cylindrical chamber,
with a length of 8.0 cm and a diameter of 1.9 cm was bored into this block. The total
working volume of the reactor is roughly 22.6 ± 0.6 cm3 and the maximum operating
pressure is estimated to be approximately 500 bar at 25 °C. Six ports were machined into
Figure 5-2. Digital photograph of the high-pressure view-cell reactor designed by
Weinstein (1998) and used for amine chemistry studies in this work.
79
the pressure vessel and were used for the inlet and outlet valves (Whitey ss-3NBM4), a
sampling valve (Valco Instrument Company, Inc. uw-type), pressure measurement, and a
thermocouple, respectively. The primary opening to the cylindrical chamber was sealed
using an ao-A120C)
3 sapphire window and a stainless steel hex nut, similar to that used for
the acoustic reactor described in Chapter 4. The pressure seal was made using a
combination of Teflon and buna-N rubber (90 durometer on the Shore A scale) gaskets.
The reactor contents can be mixed using a Teflon-coated stir bar coupled to a standard
magnetic stir plate. Pressure measurements were made using a Bourdon tube pressure
gauge (McDaniels Controls Inc.) with a readability of ±2 bar. As before, temperature was
measured to within
1
C using a T-type thermocouple (Omega Engineering). The
temperature was maintained at its set point using a PID controller (Omega Engineering
9001CN) and thermal heat tape (Barnstead).
5.3 Hetero Diels-Alder Cycloaddition
For the hetero Diels-Alder experiments, 2,3-dimethyl-1,3-butadiene was reacted
with benzylamine and formaldehyde in several different solvent systems at a temperature
of 35 °C. A general scheme for this reaction (Larson and Grieco, 1985) is
RINH2
+
HI,RI
cat. H+
-
R1
N
-
N
R2
R
(5-5)
R
R2 CHO
Reaction 5-5 shows that the amine, R 1NH 2, first reacts with the aldehyde, R 2CHO, to
form an iminium ion, R2N+HR l, which then reacts with the diene, C4 H6 , to give the
desired nitrogen heterocycle product.
80
Procedure. The temperature for these studies was chosen in order to replicate the work
of Larsen and Grieco (1985) in which they conduct this experiment in water at 35 °C for
48 h. All of the experiments conducted at atmospheric pressure were performed by Josh
Dunetz in the Danheiser laboratory. These experiments include reactions run in water,
water/tetrahydrofuran (THF), and water/hexane systems. The pressurized experiments
involving carbon dioxide were performed at an arbitrarily chosen pressure of
approximately 138 bar in the view-cell pressure vessel depicted in Figure 5-2.
A typical experiment in our lab involved loading the reactor with all the reagents
prior to sealing, heating, and pressurizing the system. Generally, specified volumes of
benzylamine,
2,3-dimethyl- 1,3-butadiene,
and aqueous formaldehyde,
which are all
liquids, were pipetted into the reactor. For runs that required more water than that present
in the aqueous formaldehyde, an excess of deionized water was added to the reaction
vessel. The reactor was then sealed using a standard torque wrench to a torque of
100 ft-lbs, and heat tape was wrapped securely around the reactor. After connecting the
reactor to the carbon dioxide system, a low flow of carbon dioxide was used to purge the
system of residual air. Next, the reactor was initially pressurized to 70 bar where it was
held for a sufficient time in order to achieve thermal equilibrium. For example, about
10 minutes were required for the reactor to reach the reaction temperature of 35 C. Once
thermal equilibrium was achieved, more carbon dioxide was added to the system until the
final pressure of 138 bar was reached. Stirring was provided by using a Teflon-coated stir
bar.
After the 48 h reaction time was completed, the carbon dioxide phase in the
reactor was vented into cold tetrahydrofuran (THF) following similar procedures to those
81
documented in Chapter 4. A high-pressure metering valve (Autoclave Engineers
10VRMM2812) heated with thermal tape (Barnstead) operated at 40 V (as set by a
standard variable transformer) was used to maintain an even flow of depressurizing
carbon dioxide and prevent formation of ice in the vent line. A glass vessel equipped with
a frit (10 m pore diameter) was used to improve recovery of any volatile components.
After the reactor was completely depressurized, it was opened and washed several times
with THF. These THF washes and the THF that the vent gas was bubbled through were
collected and given to Josh Dunetz for quantitative analysis.
Results. The Diels-Alder cycloaddition of benzylamine with 2,3-dimethyl-1,3-butadiene
was conducted in water, water/THF, water/hexane, and water/scCO
2
media at 35 °C for
48 h unless noted otherwise. The experiments involving CO 2 were conducted in this lab
while the remaining experiments were completed by Josh Dunetz. The results of these
runs are shown in Table 5-1.
Table 5-1. Results of the hetero Diels-Alder cycloaddition in various solvents. All of
these reactions were run for 48 h. The numbers in parentheses by each solvent refer to the
ratio of volumes used for each solvent in that reaction. Reaction conditions: 35 °C and
138 bar.
Dienea
Amineb
Concentration
2.2
Concentration
2.8
2.2
2.2
2.8
2.8
Solvent
Yield
H2 0
50%
H 2 0/THF (1:1)
51%
H 2 0/scCO
2 (1:
1)
0%
aDiene refers to 2,3-dimethyl-1,3-butadiene. The concentration of formaldehyde for all
experiments is equal to that of the diene. bAmine refers to benzylamine.
As shown in Table 5-1 the 50% yield obtained for the H 2 0 case is comparable to
the 64% yield of Larsen and Grieco (1985) at the same conditions. For the H 2 0/scCO
2
biphasic experiments, quantitative yields of the desired product were not achieved.
82
Instead, visual observation of the H2 0/scCO 2 reaction mixture indicated the presence of
an unexpected third phase, which appeared to be a viscous oil-like substance, below the
aqueous phase.
Analysis of this oily by-product has proven to be difficult. At this point the only
conclusions that have been reached about its identity are that the substance is a polymeric
material that is soluble only in n,n-dimethylformamide (DMF), dimethylsulfoxide
(DMSO), and 1 M HCl. Control experiments were also conducted to determine the
source of the oil formation. Table 5-2 shows the results of these visual experiments in
which one or two reagents at a time were added to a water/scCO
2
mixture at 35 °C and
138 bar. From these experiments it was concluded that the minimal reagents necessary to
form the oil are the amine and formaldehyde.
Since formaldehyde is unreactive and
benzylamine forms a white solid that is assumed to be carbamic acid when these two
reagents are added alone to the H 2 0/scCO 2 mixture, the unknown, polymeric oil is
suspected to be a product of multiple reactions between carbamic acid and formaldehyde.
Table 5-2. Results of visual experiments aimed at determining the minimal ingredients
necessary to form the unknown oil-like substance. These runs were all conducted in
H 2 0/scCO 2 mixtures at 35 °C and 138 bar at the concentrations listed in Table 5-1.
Reagents Used
Visual Observation
Diene alone
Two-phase behavior
Formaldehyde alone
Two-phase behavior
Amine alone
White solid formation
Diene and formaldehyde
Two-phase behavior
Amine and formaldehyde
Viscous oil
Amine, formaldehyde, diene
Viscous oil
83
Other possible reaction mechanisms for the formation of this oily substance are possible,
but at this point no further experiments have been pursued. Some additional control
experiments that may be of interest include changing the aldehyde to a less reactive
compound such as benzaldehyde and replacing benzylamine with a bulkier amine,
possibly a carbamate derivative, that prevents excessive carbamic acid formation.
5.4 Pictet-Spengler Cyclization
The hetero Diels-Alder experiments were abandoned after it was determined that
amine chemistry in CO 2 might be easier for transformations in which the product is
formed via an intramolecular
reaction. Reaction 5-6 shows a general scheme that
demonstrates the intramolecular cyclization that takes place during a Pictet-Spengler
reaction:
MeON
MeO
/
HCHO,
H NR
H)
H2 0
MeO
MeO
(5-6)
N. R
where the substituent R group can be varied in order to protect the nitrogen atom from
reacting with C0 2 to form carbamic acid. The hydrogen ion, H+ , is provided by a
catalytic amount of acid which was tosic acid (C7 H 80 3 S(H 2 0) or TsOH) in these
experiments.
Procedure. Unlike the hetero Diels-Alder experiments in which only one reaction
system was investigated, the Pictet-Spengler studies were more exploratory as different
amines, aldehydes, and solvents were studied. Experiments in scCO 2 were performed
using dimethoxyphenethylamine and a derivative of this compound, N-methoxycarbonyl3,4-dimethoxyphenethylamine, which replaces one of the hydrogen protons on the
84
nitrogen atom with a -CO 2Me substituent group. Besides formaldehyde, the effect of
benzaldehyde was also examined for the case when the R substituent was equal to
CO2Me. Finally, the effect of solvent was analyzed by running these reactions in toluene,
scCO 2, scCO 2/water, and neat (i.e. no solvent). The pressure of all the experiments was
set at approximately 138 bar while the temperature was either 60 or 80 °C depending on
the literature conditions that we were trying to replicate. Stirring was again provided by a
Teflon-coated stir bar.
The typical setup and procedure for these reactions was similar to that described
for the hetero Diels-Alder cycloaddition experiments. As before the reactor was initially
loaded with the reagents (the amine, aldehyde, and TsOH), purged of air with C0 2, and
pressurized to approximately 70 bar. For runs that required excess water, deionized water
was added to the system. After allowing sufficient time for the reactor contents to come
to thermal equilibrium (approximately 20 to 30 minutes depending on the reaction
temperature, 60 or 80 °C), additional CO 2 was added until the system pressure reached
138 bar. Upon completion of the reaction, the reactor was depressurized and cleaned
following the exact procedure used in the hetero Diels-Alder experiments.
Results.
The first Pictet-Spengler reaction conducted in scCO 2 was the reaction of 3,4-
dimethoxyphenethylamine
(0.5 M) with formaldehyde (0.75 M) at 60 °C and 131 bar,
reaction 5-7, where R = H for this case.
MeO
I-",
RCHO
(5-7)
MeO
NH 2
scCO 2 (1900 psi)
60 C, 19-23 h
85
Although not noted in reaction 5-7, a catalytic amount of TsOH (0.025 M) was also
added to the reaction mixture.
Unfortunately two separate runs (19 h and 23 h) of this reaction failed to yield any
quantifiable amount of the desired cyclization product depicted above. Instead it was
found that the primary amine yielded several distinct by-products. Most of these byproducts were not characterized, but one interesting product was isolated and
characterized. This by-product, the yield of which was approximately 10%, is shown
below:
MeO
MeO
MeOMN
(5-8)
'
The reaction mechanism responsible for the generation of this by-product is unknown and
further research aimed at explaining the mechanism has not been pursued.
Following the failed attempt to form the desired cyclization product starting from
a primary amine, schemes for protecting the nitrogen atom were pursued. Research by
Venkov and Lukinov (1992) on the Pictet-Spengler reaction, reaction 5-9, demonstrated
that attaching an electron withdrawing group (EWG) to the nitrogen atom gave yields
between 70% and 90% for the desired cyclization product.
MeO
MeO
~
w
H'N
EWG
RCHO
cat. TsOH-H 20
MeO
toluene,
MeO
110 °C
(59)
'EWG
(R = H,Ar)
EWG
= COR, CO 2 R,
S0 2Ph, P(O)(OEt)2
70-90%
86
Using these results, our next step was to attempt to replicate this chemistry both in
toluene and in the scCO 2 /water environment by attaching a -CO 2Me group to the primary
amine used previously. All of the experiments involving this carbamate (Nmethoxycarbonyl-3,4-dimethoxyphenethylamine), synthesized by Josh Dunetz as
described above, were run at carbamate concentrations of 0.5 M, aldehyde concentrations
(either formaldehyde or benzyaldehyde) of 0.75 M, and TsOH concentrations of 0.025 M.
The pressure was again approximately 138 bar, and the temperature was set at 60 °C for
runs using formaldehyde and 80 °C for runs that used benzaldehyde.
Table 5-3 shows the yields for complete carbamate conversion of the desired
cyclization product obtained in various solvents when formaldehyde was used as the
aldehyde source for the Pictet-Spengler cyclization of carbamate. As the results show, the
cyclization of carbamate is much faster and more efficient in toluene than in either scCO 2
or scCO 2 /H 2 0 media as it requires a shorter reaction time to reach complete conversion
(3 h versus approximately 8 h) and the yields obtained in toluene at shorter reaction times
are higher than those obtained in the alternative reaction media. In fact after 3 h only 64%
of the carbamate is even converted in the scCO 2 /H 2 0 system.
The results of reactions run with benzaldehyde, Table 5-4, as opposed to
formaldehyde support the observation that the cyclization of carbamate in toluene is
much faster than the rate that can be obtained in a scCO 2 /H2 0 mixture. In this reaction
conversion is achieved with toluene after 15 h while only 50% conversion is
100%0/o
achieved in scCO2 after 17 h.
Visual observations of the reactor contents for the experiments described in
Tables 5-3 showed that the mixture was a multi-phase system. For the case in which
87
scCO2 and H2 0 were combined in a 70:30 volume ratio, three phases were clearly present
(from top to bottom): scCO 2 , H 2 0, and a yellow oil. Although the melting point of the
carbamate, which is added to the reactor as a yellow solid, is reported to be 178-180 °C,
Table 5-3. Yields obtained in various solvents where formaldehyde was used as the
aldehyde source for the Pictet-Spengler cyclization of carbamate. The yields reported are
for reactions that were allowed enough time for complete conversion of the starting
materialc. Concentrations are: carbamate (0.5 M), formaldehyde (0.75 M), and TsOH
(0.025 M). Reaction conditions: 60 °C and 138 bar for pressurized runs.
Solvent
Reaction Time (h)
Yield
Toluene
3
99%
scCO 2 /H2 0 (70:30)a
3
64%c
scCO 2 /H2 0 (70:30)a
17
81%
scCO 2b
17
88%
scCO 2b
8
84%
aThe numbers in parentheses represent the volume ratio of scCO 2 to H2 0 in the reaction
mixture. bIn actuality there is a little water present for these experiments due to the
addition of aqueous formaldehyde to the reaction mixture. This number is actually the
conversion of the carbamate achieved after 3 h of reaction time in scC0 2 /H2 0. It is not an
actual yield of the desired product as some undesirable by-products may have formed.
Table 5-4. Conversions achieved in toluene and scCO 2 where benzaldehyde was used as
the aldehyde source for the Pictet-Spengler cyclization of carbamate. Concentrations are:
carbamate (0.5 M), benzaldehyde (0.75 M), and TsOH (0.025 M). Reaction conditions:
80 °C and 138 bar for pressurized runs.
Solvent
Reaction Time (h)
Conversion
Toluene
15
100%
scCO 2a
17
50%
aUnlike the formaldehyde case there is in fact no water present as pure benzaldehyde was
used in this run.
observations indicated that it began to melt in the reactor vessel near 50 C. In order to
determine the cause of this melting, control experiments were conducted in which the
88
carbamate was added to the reactor by itself, with only water, and with only CO 2. These
different combinations were then heated to 60 °C. For the case where the carbamate was
added by itself, nothing appeared to occur as the carbamate remained as a solid. When
water was added, the carbamate remained insoluble at low temperatures, but at
approximately 50 °C it melted and formed a second, yellow phase below that of water. A
similar observation was made for the case of pressurized CO 2 (70 bar) in which the
carbamate remained insoluble and solid at low temperatures but proceeded to melt into its
own phase near 50 °C.
From these visual observations it was hypothesized that perhaps scCO2 and water
are unnecessary for this particular reaction as they did not seem capable of solvating the
carbamate. In fact their presence may be the cause of the slower reaction rate and reduced
selectivity when compared to toluene, especially if the system is not mixed properly,
simply because they are capable of solvating the tosic acid and aldehyde but not the
carbamate, thus preventing sufficient contact between the reagents.
In order to test this hypothesis a couple of reactions were run neat (no solvents)
and compared to reactions run in toluene. Table 5-5 contains the results of these
experiments which seem to indicate that even toluene inhibits the chemical kinetics of
this particular reaction. As shown, the yields obtained in the neat experiments using
benzaldehyde
(PhCHO) and isobutyraldehyde
(i-PrCHO) as the aldehyde source are
significantly improved over those obtained in toluene at the same conditions and reaction
time. Using aqueous formaldehyde (HCHO) the yields obtained in the toluene and neat
experiments are similar.
89
Table 5-5. Yields obtained after complete conversion of the carbamate using different
aldehydes in both toluene and neat (no solvent) systems. The reaction conditions for each
experiment are shown in the table. For the case of HCHO, water was present in small
amounts due to the aqueous nature of formaldehyde.
Aldehyde
Conditions
Solvent
Yield
HCHO (aq)
60 °C, 3 h
toluene
neat
98%
100%
PhCHO
85 °C, 2 h
toluene
neat
21%
81%
i-PrCHO
80 °C, 23 h
toluene
neat
39%
77%
Discussion. The results presented above suggest that scCO2 and water do not provide
any technological advantages for this particular Pictet-Spengler reaction system as the
chemical kinetics of this reaction are much slower and less selective in these alternative
reaction media than in toluene. However, one must keep in mind that mixing in our
reactor was not ideal as the stir bar was incapable of vigorously mixing all three phases
even at its maximum rotational speed. Also we were not able to close mass balances as
the carbamate had a tendency to solidify in the outlet lines upon reactor venting.
Although we were able to recover approximately
90% of the carbamate in several
practice runs, it is possible that some of the desired product could be lost during the
venting and cleaning process. Finally, with the exception of using formaldehyde as the
aldehyde source, toluene, even with its high solubility for the reagents used in this
reaction, does not offer any advantage over running the reaction neat. This result suggests
that solvents as a reaction promoter may be unnecessary, which further supports the
conclusion that this scCO2/water biphasic mixture will not provide any technological
90
advantages since it is incapable of solvating the carbamate reactant that was used in this
study.
5.5 Carbamate Generation Using CO 2 as a Carbon Source
Although the Pictet-Spengler reaction proved to be disappointing from the
standpoint of potentially using water and carbon dioxide as alternative, green solvents, it
has led us in a new direction in which carbon dioxide can be used as both a reactant and
solvent. The plan for this new path is to basically conduct a "one-pot" synthesis in which
carbamate compounds
are first generated using CO 2 as a reactant.
Generating the
carbamate in this manner would allow for the elimination of several organic compounds,
such as dichloromethane, that are required for making the carbamate using conventional
techniques. The second step would then involve injection of additional reagents required
for synthetic transformations
of interest, such as the Pictet-Spengler
reaction, that are
capable of producing nitrogen heterocycles from carbamate starting material.
The fortnation of carbamates from primary amines and CO2 has been well
documented in the literature (Salvatore et al., 2002; Selva et al., 2002; Yoshida et al.,
2000; McGhee et al., 1995). Many different techniques have been utilized by these
studies; one of which we are attempting to replicate (Yoshida et al., 2000) and another
that we plan to replicate (Selva et al., 2002). The chemical equations for these two
methods were shown previously in reactions 5-2 and 5-3. The overall goal of this portion
of the project was to find or improve upon a method for generating carbamate products in
good yields from primary amines and CO 2 with the added constraint that the reagents
used in this method should be compatible with all reagents used in any seceding synthetic
91
transformations. To date we have only attempted to replicate the Yoshida group's (2000)
reactions with unsatisfactory results.
Procedure.
The procedures that were used to load and empty the reactor are similar to
those discussed previously in this chapter. The reactor is first loaded with all the reagents
which for this case include: benzylamine (0.10 M), butyl bromide (BuBr) (0.16 M),
tetrabutylammonium
bromide (5 mol%), and potassium carbonate (0.2 M). Because the
butyl bromide can potentially cause undesired N-alkylation of the starting amine, care
was taken to prevent
their contact by
loading in order the benzylamine,
tetrabutylammonium bromide, potassium carbonate, and finally the butyl bromide. The
reactor was then quickly sealed at a torque of 150 ft-lbs, wrapped with heat tape, and
pressurized to approximately 48 bar. After allowing 30 minutes for the reactor to reach
thermal equilibrium at 100
C, additional CO 2 was added to bring the final system
pressure to 103 bar. Stirring was provided by a Teflon-coated stir bar.
After 5 h the reaction was quenched by bubbling the CO2 phase through cold THF
using the equipment described earlier. Once depressurized the vessel was opened and
washed several times with THF and water. These washes and the cold THF were
collected and given to Josh Dunetz for analysis.
Results. Our experimental results have been disappointing as we have been unable to
replicate the isolated yield of 85% obtained by the Yoshida group (2000). Table 5-6
shows that when the reactor is in the vertical position we were only able to convert 40%
of our starting material into the desired carbamate product with another 10% and 35%
converted into two undesired products that form as a result of N-alkylation (described
92
below). One encouraging result is that less than 5% of the starting benzylamine remained
unreacted.
Table 5-6. Yields obtained from the carbamate generation experiments in two reactor
orientations: vertical and horizontal. The listed compounds are benzylamine (1), the
desired carbamate (2), and the undesired N-alkylation by-products, (3) and (4).
Reactor Orientation
Chemical Compounds
Vertical
Horizontal
(1)
<5%
34%
11H3
(2)
40%
23%
13,
(3)
10%
6%
(4)
35%
37%
¢-E H,
N
N
(
CH3
3
113
H
A possible cause of these less than desirable results is poor mixing conditions that
may be present in the reactor. Visual observation of the reactor contents seems to indicate
that before addition of CO2 the mixture of reagents appears to be a thick slurry that is
capable of being stirred with a stir bar. However, upon addition of CO 2 the mixture
becomes very viscous and stirring ceases after the stir bar becomes trapped in the reaction
mixture. Eventually, the stir bar does free itself but the degree of mixing never appears
sufficient enough to obtain good contact between the multiple phases that are present.
One solution to improve the mixing conditions was to run the reactor in a horizontal
93
position in order to increase the contact area between the CO 2 phase and the other
reagents. However, this solution, which we expected to provide better mixing, actually
produced yields worse than those observed when the reactor was in the vertical position.
These results represent only one experimental run in each reactor configuration so it is
likely that the results reported for the horizontal and/or vertical configurations may be
incorrect. Further experiments to study the reproducibility of the results for each
configuration are planned. As shown in Table 5-6 the final workup in the horizontal
position contained only 23% of the desired product, 6% and 37% of the two undesired Nalkylation products, and 34% of unreacted starting amine.
Discussion.
The large extent to which N-alkylation is occurring in our system is a major
concern. From Figure 5-3, which is the mechanism for carbamate generation using the
Yoshida group's methodology, N-alkylation (although not depicted) is the direct reaction
of butyl bromide (R3 -X) with the nitrogen atom in benzylamine (R1R2NH). This reaction
can occur twice with the same nitrogen atom to form compound 4 depicted in Table 5-6
or it can occur once as shown in compound 3. The other type of alkylation, O-alkylation,
is the reaction of butyl bromide (R 3-X) with the carbamate anion as depicted in Figure 53. O-alkylation leads to the desired product, compound 2 in Table 5-6, and it is also the
source of the -OBu group on compound 3.
94
K2CO
3
e
Br
®
(
2
H2 NR 1 R
00
,Nn'n'NBu0 4
I`NN
2RRnZNH
12 RR2NH
-.
.~f-
CO2
,
R
'N
i2
2
0 H2NRR2
Bu 4 NBr
R
NBU4
>N
12
R3 -X
0
R3
R1 N .
N
R2
R
Bu4N X
Figure 5-3. Mechanism for carbamate generation using scCO 2 as a carbon source for the
reaction scheme described by the Yoshida group (2000). Only the primary pathway for
generation of the desired carbamate compound is shown. The pathways for undesired byproduct formation via N-alkylation reactions are not depicted. For the reaction scheme
that we investigated: R = CH2 C6 H5; R2 = H; R3 = C4H9 . Potassium carbonate is also
present, and its role is to react with the Bu 4N+X- salt to form KX-; thus regenerating the
phase-transfer catalyst, Bu4 N+.
The Yoshida group with a reactor setup similar to ours (50 mL, cylindrical, 316stainless steel, and stir bar) does not observe N-alkylation. In general the ratio of N-
alkylation versus the desired O-alkylation will depend on a couple of parameters. First,
the amount of carbamate anion, BnNHCO2- Z+ (where Z is the counter ion such as NBu 4
and K+), that is present has a large effect since the more the amine is sequestered as the
carbamate anion, the less the amount of undesirable N-alkylation that can occur. Because
the carbamate anion can be in equilibrium with the amine and CO2, one possible way to
generate more carbamate anions is to adjust the system pressure in order to drive the
equilibrium reaction toward the production of carbamate anions. A second possible cause
of N-alkylation is the inability of the carbamate anion and the alkylating agent (BuBr),
which respectively reside in the liquid slurry and the scCO 2 phase, to react due to poor
95
mixing between the different phases within the reactor; thus slowing the rate of the
desired O-alkylation reaction but allowing more time for N-alkylation to occur.
5.6 Conclusions and Recommendations
Our work so far has clearly demonstrated that conducting nitrogen-bearing
chemistry in the presence of carbon dioxide is very challenging. All of the reaction
systems that we have studied contain complications, including both the feasibility of
certain chemical pathways in scCO 2 and physical issues such as the degree of mixing in
our reactor apparatus, which have inhibited our progress. Despite the problems that we
have encountered, a few conclusions can be drawn:
·
When using scCO2 as a solvent, primary amines are not a good starting
material due to their reactivity with carbon dioxide. Both the hetero DielsAlder and Pictet-Spengler methods were incapable of using primary
amines.
*
Carbamates
as a starting material offer promise for use in synthetic
transformations in the presence of CO2. Reasonable yields (84% in scCO2
versus 100% in toluene after 8h) and conversions (64% in scCO2 versus
100% in toluene after 3 h) were obtained for the Pictet-Spengler reaction
when starting with the carbamate, N-methoxycarbonyl-3,4-dimethoxyphenethylamine. However, the advantage of using carbon dioxide, water,
or even toluene as solvents for this particular reaction is debatable as the
reaction proceeded significantly faster when run neat (i.e. with no
solvents).
96
* Carbamates would be even more attractive as a protecting group for
nitrogen if a "one-pot" synthesis can be developed to generate the desired
nitrogen heterocycle from a primary amine via a carbamate intermediate.
Because the role of carbon dioxide and water in the Pictet-Spengler
reaction involving the carbamate is questionable, the generation of
carbamates using CO2 as both a reactant and solvent would give carbon
dioxide a quantifiable use as it would replace conventional solvents such
as phosgene which are normally used for the synthesis of carbamates.
Despite our failed attempts to replicate the chemistry of the Yoshida
group, this methodology
is still encouraging
as our collaborators
at
Cambridge University, UK have been able to obtain an isolated yield of
76% for the desired product using a reactor setup similar to ours.
As for recommendations, there are a few studies that would be beneficial for the
completion of the project's objectives:
* The Selva group's (2002) procedure for generating carbamates from CO2
should be investigated and replicated. This reaction scheme is intriguing
as it uses dimethyl carbonate (DMC), which is more "green" than BuBr, as
an alkylating agent, and it does not require any added base or catalysts.
DMC also may act as a beneficial co-solvent for further desired reactions
such as the Pictet-Spengler reaction.
* Attempts to replicate the chemistry of the Yoshida group (2000) should be
continued. Our failure to obtain yields reasonably
close to those of
Yoshida and co-workers are likely due to inefficient mixing and/or poor
97
temperature control in our reactor system. Procurement of a new autoclave
reactor with better agitation, temperature control, and a variable feed
injection system is underway.
*
)nce a successful method has been developed for the generation of
carbamates from C0 2, "one-pot" methods for the synthesis of nitrogen
heterocycles should be investigated. Pictet-Spengler cyclizations, hetero
Diels-Alder cycloadditions, and dipolar cycloadditions among others are
synthetic transformations that can be studied.
5.7 References
Belli Dell'Amico, D., F. Calderazzo, L. Labella, F. Marchetti and G. Pampaloni,
"Converting Carbon Dioxide into Carbamate Derivatives." Chem. Rev. 2003, 103,
857.
Breslow, R., U. Maitra and D. Rideout, "Selective Diels-Alder Reactions in Aqueous
Solutions and Suspensions." Tetrahedron Lett. 1983, 24, 1901.
Chandrasekhar, J., S. Shariffskul and W.L. Jorgensen, "QM/MM Simulations for DielsAlder Reactions in Water: Contribution of Enhanced Hydrogen Bonding at the
Transition State to the Solvent Effect." J. Phys. Chem. B 2002, 106, 8078.
Cox, E. D. and J.M. Cook, "The Pictet-Spengler Condensation: A New Direction for an
Old Reaction." Chem. Rev. 1995, 95, 1797.
Fischer, H., O. Gyllenhaal, J. Vessman and K. Albert, "Reaction Monitoring of Aliphatic
Amines in Supercritical Carbon Dioxide by Proton Nuclear Magnetic Resonance
Spectroscopy and Implications for Supercritical Fluid Chromatography." Anal.
Chem. 2003, 75, 622.
Glebov, E.M., L.G. Krishtopa, V. Stepanov and L.N. Krasnoperov, "Kinetics of a Diels-
Alder Reaction of Maleic Anhydride and Isoprene in Supercritical CO2." J. Phys.
Chem. A 2001, 105, 9427.
Gothelf, K. V. and K.A. Jorgensen, "Asymmetric 1,3-Dipolar Cycloaddition Reactions."
Chem. Rev. 1998, 98, 863.
98
Grieco, P. A., K. Yoshida, and Z. He, "Sodium 6-methoxy-(E,E) -3,5-hexadienoate: A
Useful Diene in the Aqueous Diels-Alder Reaction." Tetrahedron Lett. 1984, 25,
5715.
Grieco, P.A., P. Garner and Z. He, "Micellar Catalysis in the Aqueous Intermolecular
Diels-Alder Reaction: Rate Acceleration and Enhanced Selectivity." Tetrahedron.
Lett. 1983, 24, 1897.
Jacobson, G.B., C.T. Lee, R.P. daRocha and K.P. Johnston, "Organic Synthesis in
Water/Carbon Dioxide Emulsions." J. Org. Chem. 1999, 64, 1207.
Kainz, S., A. Brinkmann, W. Leitner and A. Pfaltz, "Iridium-Catalyzed Enantioselective
Hydrogenation of Imines in Supercritical Carbon Dioxide." J. Am. Chem. Soc.
1999, 121, 6421.
Kumar, P., K.N. Dhawan, K. Kishor, K.P. Bhargava and R.K. Satsangi, "1-Phenyl-6,7disubstituted 2-(3-disubstituted aminopropyl)-1,2,3,4-tetrahydroisoquinolines as
Possible Antitubercular Agents." J. Heterocycl. Chem. 1982, 19, 677.
Larsen, S.D. and P.A. Grieco, "Aza Diels-Alder Reactions in Aqueous Solution:
Cyclocondensation of Dienes with Simple Iminium Salts Generated Under
Mannich Conditions." J. Am. Chem. Soc. 1985, 107, 1768.
Leitner, W., "Supercritical Carbon Dioxide as a Green Reaction Medium for Catalysis."
Acc. Chem. Res. 2002, 35, 746.
Lubineau, A. and J. Auge, "Water as a Solvent in Organic Synthesis." Topics in Current
Chemistry, 1999, 206, 1.
McCarthy, M., H. Stemmer and W. Leitner, "Catalysis in Inverted Supercritical
CO2/Aqueous Biphasic Media." Green Chem. 2002, 4, 501.
McGhee, W., D. Riley, K. Christ, Y. Pan and B. Parnas, "Carbon Dioxide as a Phosgene
Replacement: Synthesis and Mechanistic Studies of Urethanes from Amines,
CO 2, and Alkyl Chlorides." J. Org. Chem. 1995, 60, 2820.
Otto, S. and J.B.F.N. Engberts, "Diels-Alder Reactions in Water." Pure Appl. Chem.
2000, 72, 1365.
Qian, J., M.T. Timko, A.J. Allen, C.J. Russell, B. Winnik, B. Buckley, J.I. Steinfeld and
J.W. Tester, "Solvophobic Acceleration of Diels-Alder Reactions in Supercritical
Carbon Dioxide." J. Am. Chem. Soc. 2004, 126, 5465.
Renslo, A. R., R.D. Weinstein, J.W. Tester and R.L. Danheiser, "Concerning the
Regiochemical Course of the Diels-Alder Reaction in Supercritical Carbon
Dioxide." J. Org. Chem. 1997, 62, 4530.
99
Rideout, D.C. and R. Breslow, "Hydrophobic Acceleration of Diels-Alder Reactions." J.
Am. Chem. Soc., 1980, 102, 7816.
Salvatore, R.N., F. Chu, A.S. Nagle, E.A. Kapxhiu, R.M. Cross and K.W. Jung,
"Efficient Cs2CO3 -promoted Solution and Solid Phase Synthesis of Carbonates
and Carbamates in the Presence of TBAI." Tetrahedron 2002, 58, 3329.
Selva, M., P. Tundo and A. Perosa, "The Synthesis of Alkyl Carbamates from Primary
Aliphatic Amines and Dialkyl Carbonates in Supercritical Carbon Dioxide."
Tetrahedron Lett. 2002, 43, 1217.
Soerens, D., J. Sandrin, F. Ungemach, P. Mokry, G.S. Wu, E. Yamanaka, L. Hutchins,
M. DiPierro and J.M. Cook, "Study of the Pictet-Spengler Reaction in Aprotic
Media: Synthesis of the f-galactosidase Inhibitor, Pyridindolol." J. Org. Chem.
1979, 44, 535.
Thompson, R.L., R. Glaser, D. Bush, C.L. Liotta and C.A. Eckert, "Rate Variations of a
Hetero-Diels-Alder Reaction in Supercritical Fluid CO2 ." Ind. Eng. Chem. Res.
1999, 38, 4220.
Timko, M.T., "Acoustic Emulsions of Liquid, Near-Critical Carbon Dioxide and Water:
Application to Synthetic Chemistry Through Reaction Engineering." Ph.D.
Thesis, Department of Chemical Engineering, Massachusetts Institute of
Technology, Cambridge, MA, 2004.
Venkov, A. P. and L.K. Lukanov, "Improved Methods for the Synthesis of NAcyltetrahydroisoquinolines." Syn. Comm. 1992, 22, 3235.
Weinstein, R.W., "Organic Synthesis in Supercritical Carbon Dioxide." Ph.D. Thesis,
Department of Chemical Engineering, Massachusetts Institute of Technology,
Cambridge, MA, 1998.
Yokoyama, A., T. Ohwada and K. Shudo, "Prototype Pictet-Spengler Reactions
Catalyzed by Superacids. Involvement of Dicationic Superelectrophiles." J. Org.
Chem. 1999, 64, 611.
Yoshida, M., N. Hara and S. Okuyama, "Catalytic Production of Urethanes from Amines
and Alkyl Halides in Supercritical Carbon Dioxide, Chem. Comm. 2000, 151.
100
CHAPTER 6
CONCLUSIONS AND RECCOMMENDATIONS
The overall objective of this work was to investigate reaction systems in which
dense carbon dioxide and water were used as "green" solvents in place of conventional,
organic compounds. This research successfully demonstrated carbon dioxide's ability to
accelerate certain Diels-Alder reactions via a solvophobic mechanism. It also followed up
on the work of Michael Timko in demonstrating the ability of power ultrasound to
accelerate a Diels-Alder reaction in a liquid water/dense carbon dioxide biphasic system
through the generation of emulsions. Finally, this thesis began initial work in developing
reaction systems capable of producing nitrogen heterocycles in the presence of CO2.
These three studies clearly demonstrated the technological advantages that can be
gained by selectively using water and/or scCO2 for certain reaction systems such as those
investigated in the solvophobic and ultrasonic acceleration studies. However, nitrogen
chemistry proved to be much more complicated in the presence of CO2 because of the
reactivity of amines with CO2 to form an undesired by-product called carbamic acid.
Future work aimed at demonstrating the technological advantages of CO2 and/or water as
alternative reaction media basically involve further quantification of both carbon
dioxide's and water's role in accelerating chemical reactions and continued investigation
of amine chemistry in the presence of CO2.
Below are specific conclusions and recommendations for each of the three studies
that were conducted as part of this thesis:
101
Solvophobic Acceleration of a Diels-Alder Reaction in Supercritical Carbon Dioxide
The Diels-Alder reaction of N-ethylmaleimide with 9-hydroxymethylanthracene
was found to proceed at rates in supercritical carbon dioxide that are much faster than in
traditional
organic
solvents.
On
the
basis
of
the
low
solubility
of
9-
hydroxymethylanthracene in scCO2, a solvophobic mechanism, consistent with that
proposed for acceleration of this reaction in water and fluorocarbon solvents, was
hypothesized
to be source of the accelerated kinetics in scCO 2. This hypothesis is
supported by the observed negative pressure and density dependencies of the rate
constant which are driven by the positive relationship between fluid density and solute
solubility. Clustering phenomena (solute/solvent or solute/solute) arising from density
fluctuations near the critical point were ruled out as a potential source of the observed
kinetic acceleration because the apparent activation volumes are both large and positive
(+350 cm3 mol 1) and only a weak function of reduced temperature. Instead, the large
activation volumes can be attributed to changes in the solubility of the reagents relative to
that of the transition state with increasing density.
Understanding the solvophobic acceleration of Diels-Alder reactions in scCO2
provides a tool for selection of model reactions to conduct in supercritical fluids.
Typically, reagent selection is based on solubility in scCO2, but our results show that
solvophobic acceleration can provide a second criterion for the choice of reagents. The
similarities of scCO2 with fluorinated solvents (which are more accessible experimentally
than scCO2) might also be exploited in the future. However, like water, scale-up of
reactions involving sparingly soluble species may be prohibitive for utilizing scCO2 as a
solvent for Diels-Alder reactions and other syntheses.
102
Future work should be conducted to further understand this solvophobic
acceleration so that the results uncovered by this investigation might be applied to a
wider range of organic reactions. Possible focal points include: (1) determination of
specific molecular structures which are expected to interact solvophobically in scCO 2, (2)
further quantification of the solvophobic effect, particularly in the presence of hydrogenbond-donating co-solvents, and (3) development of computational techniques integrating
methods from density functional theory and molecular simulation to further molecularlevel understanding of reactivity in scCO2.
Water as a Catalyst: Conversion and Selectivity of a Diels-Alder Reaction in Carbon
Dioxide/Water Systems
A set of experiments were conducted at a water volume fraction of 0.85 while
maintaining the optimal horn-to-interface distance of 2 cm and varying the initial
concentration of methyl vinyl ketone from 0.05 to 1.0 mol L- 1. Under sonicated
conditions the most dramatic improvements were seen when the initial methyl vinyl
ketone concentration was 1 mol L- . At this concentration the endo:exo selectivity was
nearly 13:1 and the conversion after one hour was greater than 90%. These results
compare favorably to that obtained silently (6:1 selectivity, 70% conversion) or expected
in pure carbon dioxide (4.5:1 selectivity, 30% conversion). As the concentration of
methyl vinyl ketone was decreased from 1.0 to 0.05 mol L - , the selectivities and
conversions obtained under sonicated conditions approached those obtained without
ultrasonic emulsification (16:1 selectivity, 50% conversion). It was concluded that for
systems in which the water-based Hatta number is greater than one, sonication increases
mass transport rates in this biphasic system, thus allowing both the selectivity and
103
conversion to approach that of water while avoiding some of the solubility limitations of
water.
One possibility for future work in the generation of emulsions via power
ultrasound is the design of large-scale reactors capable of efficiently producing
ultrasonically-induced emulsions. Designs that use vibrating walls or recycle loops that
contain a sonicated zone are possible solutions that might allow for technology transfer
from the laboratory scale to the industrial scale.
Synthesis of Nitrogen-Bearing Compounds in scCO 2 and/or H 2 0 Systems
Conducting nitrogen-bearing chemistry in the presence of carbon dioxide was
found to be difficult as the reaction systems that we studied were rife with complications
that have inhibited our progress. Despite the problems that we encountered,
a few
conclusions were drawn:
*
Carbamates
as a starting material offer promise for use in synthetic
transformations
in the
presence
of
CO 2.
Reasonable
yields
and
conversions were obtained for the Pictet-Spengler reaction when starting
with the carbamate, N-methoxycarbonyl-3,4-dimethoxyphenethylamine.
However, the purpose of using carbon dioxide, water, and even toluene as
solvents for this particular reaction is debatable as the reaction proceeded
significantly faster when run neat (i.e. no solvents).
*
Carbamates would be even more attractive as a protecting group for
nitrogen if a "one-pot" synthesis can be developed to generate the desired
nitrogen heterocycle from a primary amine via a carbamate intermediate.
Because the role of carbon dioxide and water in the Pictet-Spengler
104
reaction involving the carbamate is questionable, the generation of
carbamates using CO2 as both a reactant and solvent would give carbon
dioxide a quantifiable use as it would replace conventional solvents such
as phosgene which are normally used for the synthesis of carbamates. The
results of Yoshida (2000) are encouraging despite our failed attempts to
replicate this chemistry.
As for recommendations, there are a few studies that would be beneficial for the
completion of the project's objectives:
·
he Selva (2002) procedure for generating carbamates from CO 2 should
be investigated and replicated. This reaction scheme is intriguing as it uses
dimethyl carbonate (DMC), which is more "green" than BuBr, as an
alkylating agent, and it does not require any added base or catalysts. DMC
also may act as a beneficial co-solvent for further desired reactions such as
the Pictet-Spengler reaction.
* Attempts to replicate the chemistry of Yoshida (2000) should be
continued. Our failure to obtain yields reasonably
close to those of
Yoshida is likely due to inefficient mixing in our reactor system. Plans to
buy a commercial reactor with better mixing conditions are currently in
the early stages.
* Once a successful method has been developed for the generation of
carbamates from CO2, "one-pot" methods for the synthesis of nitrogen
heterocycles should be investigated. Pictet-Spengler cyclizations, hetero
105
Diels-Alder cycloadditions, and dipolar cycloadditions among others are
synthetic transformations that can be studied.
106
CHAPTER 7
APPENDICES
7.1 Experimental Data: Solvophobic Acceleration
Data for the 75 C runs
The reaction rates for the other temperatures are reported in Table 3-1.
P (bar)
p (kg m3' )
[A]a
(106 M)
[Blb
(10 4 M)
pseudo-first-
R2
kc
(10 3 L mol
order-slope
-
(105 s 1)
128
352 + 8
5.01
5.00
14.4
0.98
290
138
406
9
9.95
10.2
23.4
0.98
230
152
472 3 0
4.28
4.26
5.14
0.99
120
169
540
8
6.99
6.94
7.75
0.99
110
179
572
7
5.01
5.00
3.79
0.98
76
193
610
22
8.02
8.02
5.64
0.98
70
a[A] = concentration of 9-hydroxymethylanthracene
b[B] = concentration of N-ethylmaleimide
data file: 75 C solvophobic data.xls
'1
s)
107
7.2 Experimental Data: Diels-Alder Conversion and Selectivity
Data reported in Chapter 4
Effect of Initial Concentration of methyl vinyl ketone:
Dw= 0.85; 30 °C; 80 bar; 0.6 W cm-3; 20 kHz; 25% duty cycle; 1 hr
Silent Conditions
[MVK]o (mol L 1)
X
S = NIX
0.42
0.37
0.47
0.53
0.63
0.69
0.69
16.2
13.3
9.8
10.4
6.8
6.4
6.1
[MVK]o (mol L-)
X
S = NIX
0.05
0.05
0.20
0.50
1.00
0.53
NA
0.77
0.92
0.91
15.8
15.7
14.5
14.6
12.5
0.05
0.10
0.35
0.40
0.70
0.94
1.00
Acoustic Conditions